Skip to content

P
PaddleDetection

PaddlePaddle / PaddleDetection
1 年多 前同步成功

通知 696
Star 11112
Fork 2696
P
PaddleDetection
未验证 提交 a62f6803 编写于 作者: F FDInSky 提交者： GitHub
浏览文件
操作

clear directories & files for dygraph branch (#829) 

clear directories & files for dygraph branch
上级 7cf31190
展开全部 隐藏空白更改
内联 并排
Showing with 1 addition and 44780 deletion
.travis.yml
浏览文件 @ a62f6803
......@@ -20,7 +20,7 @@ addons:
before_install:
- sudo pip install -U virtualenv pre-commit pip
- docker pull paddlepaddle/paddle:latest
- git pull https://github.com/PaddlePaddle/PaddleDetection master -r
- git pull https://github.com/PaddlePaddle/PaddleDetection dygraph -r
script:
- exit_code=0
......
README_en.md 已删除 100644 → 0
浏览文件 @ 7cf31190
English | [简体中文](README.md)
Documentation:[https://paddledetection.readthedocs.io](https://paddledetection.readthedocs.io)
# PaddleDetection
PaddleDetection is an end-to-end object detection development kit based on PaddlePaddle, which
aims to help developers in the whole development of training models, optimizing performance and
inference speed, and deploying models. PaddleDetection provides varied object detection architectures
in modular design, and wealthy data augmentation methods, network components, loss functions, etc.
PaddleDetection supported practical projects such as industrial quality inspection, remote sensing
image object detection, and automatic inspection with its practical features such as model compression
and multi-platform deployment.
**Now all models in PaddleDetection require PaddlePaddle version 1.7 or higher, or suitable develop version.**
<div align="center">
<img src="docs/images/000000570688.jpg" />
</div>
## Introduction
Features:
- Rich models:
PaddleDetection provides rich of models, including 100+ pre-trained models
such as object detection, instance segmentation, face detection etc. It covers
the champion models, the practical detection models for cloud and edge device.
- Production Ready:
Key operations are implemented in C++ and CUDA, together with PaddlePaddle's
highly efficient inference engine, enables easy deployment in server environments.
- Highly Flexible:
Components are designed to be modular. Model architectures, as well as data
preprocess pipelines, can be easily customized with simple configuration
changes.
- Performance Optimized:
With the help of the underlying PaddlePaddle framework, faster training and
reduced GPU memory footprint is achieved. Notably, YOLOv3 training is
much faster compared to other frameworks. Another example is Mask-RCNN
(ResNet50), we managed to fit up to 4 images per GPU (Tesla V100 16GB) during
multi-GPU training.
Supported Architectures:
| | ResNet | ResNet-vd <sup>[1](#vd)</sup> | ResNeXt-vd | SENet | MobileNet | HRNet | Res2Net |
| ------------------- | :----: | ----------------------------: | :--------: | :---: | :-------: |:------:|:-----: |
| Faster R-CNN | ✓ | ✓ | x | ✓ | ✗ | ✗ | ✗ |
| Faster R-CNN + FPN | ✓ | ✓ | ✓ | ✓ | ✗ | ✓ | ✓ |
| Mask R-CNN | ✓ | ✓ | x | ✓ | ✗ | ✗ | ✗ |
| Mask R-CNN + FPN | ✓ | ✓ | ✓ | ✓ | ✗ | ✗ | ✓ |
| Cascade Faster-RCNN | ✓ | ✓ | ✓ | ✗ | ✗ | ✗ | ✗ |
| Cascade Mask-RCNN | ✓ | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ |
| Libra R-CNN | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
| RetinaNet | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| YOLOv3 | ✓ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
| SSD | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
| BlazeFace | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
| Faceboxes | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
<a name="vd">[1]</a> [ResNet-vd](https://arxiv.org/pdf/1812.01187) models offer much improved accuracy with negligible performance cost.
More models:
- EfficientDet
- FCOS
- CornerNet-Squeeze
- YOLOv4
More Backbones:
- DarkNet
- VGG
- GCNet
- CBNet
Advanced Features:
- [x] **Synchronized Batch Norm**
- [x] **Group Norm**
- [x] **Modulated Deformable Convolution**
- [x] **Deformable PSRoI Pooling**
- [x] **Non-local and GCNet**
**NOTE:** Synchronized batch normalization can only be used on multiple GPU devices, can not be used on CPU devices or single GPU device.
The following is the relationship between COCO mAP and FPS on Tesla V100 of representative models of each architectures and backbones.
<div align="center">
<img src="docs/images/map_fps.png" />
</div>
**NOTE:**
- `CBResNet` stands for `Cascade-Faster-RCNN-CBResNet200vd-FPN`, which has highest mAP on COCO as 53.3% in PaddleDetection models
- `Cascade-Faster-RCNN` stands for `Cascade-Faster-RCNN-ResNet50vd-DCN`, which has been optimized to 20 FPS inference speed when COCO mAP as 47.8%
- The enhanced `YOLOv3-ResNet50vd-DCN` is 10.6 absolute percentage points higher than paper on COCO mAP, and inference speed is nearly 70% faster than the darknet framework
- All these models can be get in [Model Zoo](#Model-Zoo)
## Tutorials
### Get Started
- [Installation guide](docs/tutorials/INSTALL.md)
- [Quick start on small dataset](docs/tutorials/QUICK_STARTED.md)
- [Train/Evaluation/Inference](docs/tutorials/GETTING_STARTED.md)
- [FAQ](docs/FAQ.md)
### Advanced Tutorial
- [Guide to preprocess pipeline and custom dataset](docs/advanced_tutorials/READER.md)
- [Models technical](docs/advanced_tutorials/MODEL_TECHNICAL.md)
- [Transfer learning document](docs/advanced_tutorials/TRANSFER_LEARNING.md)
- [Parameter configuration](docs/advanced_tutorials/config_doc):
- [Introduction to the configuration workflow](docs/advanced_tutorials/config_doc/CONFIG.md)
- [Parameter configuration for RCNN model](docs/advanced_tutorials/config_doc/RCNN_PARAMS_DOC.md)
- [IPython Notebook demo](demo/mask_rcnn_demo.ipynb)
- [Model compression](slim)
- [Model compression benchmark](slim)
- [Quantization](slim/quantization)
- [Model pruning](slim/prune)
- [Model distillation](slim/distillation)
- [Neural Architecture Search](slim/nas)
- [Deployment](deploy)
- [Export model for inference](docs/advanced_tutorials/deploy/EXPORT_MODEL.md)
- [Python inference](deploy/python)
- [C++ inference](deploy/cpp)
- [Inference benchmark](docs/advanced_tutorials/inference/BENCHMARK_INFER_cn.md)
## Model Zoo
- Pretrained models are available in the [PaddleDetection model zoo](docs/MODEL_ZOO.md).
- [Mobile models](configs/mobile/README.md)
- [Anchor free models](configs/anchor_free/README.md)
- [Face detection models](docs/featured_model/FACE_DETECTION_en.md)
- [Pretrained models for pedestrian detection](docs/featured_model/CONTRIB.md)
- [Pretrained models for vehicle detection](docs/featured_model/CONTRIB.md)
- [YOLOv3 enhanced model](docs/featured_model/YOLOv3_ENHANCEMENT.md): Compared to MAP of 33.0% in paper, enhanced YOLOv3 reaches the MAP of 43.6%, and inference speed is improved as well
- [Objects365 2019 Challenge champion model](docs/featured_model/champion_model/CACascadeRCNN.md)
- [Best single model of Open Images 2019-Object Detction](docs/featured_model/champion_model/OIDV5_BASELINE_MODEL.md)
- [Practical Server-side detection method](configs/rcnn_enhance/README_en.md): Inference speed on single V100 GPU can reach 20FPS when COCO mAP is 47.8%.
## License
PaddleDetection is released under the [Apache 2.0 license](LICENSE).
## Updates
v0.3.0 was released at `05/2020`, add anchor-free, EfficientDet, YOLOv4, etc. Launched mobile and server-side practical and efficient multiple models. For example, the YOLOv3-MobileNetv3 mobile side model is accelerated 3.5 times, the server side has optimized the two-stage model, and the speed and accuracy have high cost performance. We also refactored predictive deployment functions, and improved ease of use, fix many known bugs, etc.
Please refer to [版本更新文档](docs/CHANGELOG.md) for details.
## Contributing
Contributions are highly welcomed and we would really appreciate your feedback!!
configs/anchor_free/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
# Anchor Free系列模型
## 内容
- [简介](#简介)
- [模型库与基线](#模型库与基线)
- [算法细节](#算法细节)
- [如何贡献代码](#如何贡献代码)
## 简介
目前主流的检测算法大体分为两类： single-stage和two-stage，其中single-stage的经典算法包括SSD, YOLO等，two-stage方法有RCNN系列模型，两大类算法在[PaddleDetection Model Zoo](../../docs/MODEL_ZOO.md)中均有给出，它们的共同特点是先定义一系列密集的，大小不等的anchor区域，再基于这些先验区域进行分类和回归，这种方式极大的受限于anchor自身的设计。随着CornerNet的提出，涌现了多种anchor free方法，PaddleDetection也集成了一系列anchor free算法。
## 模型库与基线
下表中展示了PaddleDetection当前支持的网络结构，具体细节请参考[算法细节](#算法细节)。
| | ResNet50 | ResNet50-vd | Hourglass104 |
|:------------------------:|:--------:|:--------------------------:|:------------------------:|
| [CornerNet-Squeeze](#CornerNet-Squeeze) | x | ✓ | ✓ |
| [FCOS](#FCOS) | ✓ | x | x |
### 模型库
#### COCO数据集上的mAP
| 网络结构 | 骨干网络 | 图片个数/GPU | 预训练模型 | mAP | FPS | 模型下载 | 配置文件 |
|:------------:|:--------:|:----:|:-------:|:-------:|:---------:|:----------:|:----------:|
| CornerNet-Squeeze | Hourglass104 | 14 | 无 | 34.5 | 35.5 | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/cornernet_squeeze_hg104.tar) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/anchor_free/cornernet_squeeze_hg104.yml) |
| CornerNet-Squeeze | ResNet50-vd | 14 | [faster\_rcnn\_r50\_vd\_fpn\_2x](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_vd_fpn_2x.tar) | 32.7 | 42.45 | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/cornernet_squeeze_r50_vd_fpn.tar) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/anchor_free/cornernet_squeeze_r50_vd_fpn.yml) |
| CornerNet-Squeeze-dcn | ResNet50-vd | 14 | [faster\_rcnn\_dcn\_r50\_vd\_fpn\_2x](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_dcn_r50_vd_fpn_2x.tar) | 34.9 | 40.05 | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/cornernet_squeeze_dcn_r50_vd_fpn.tar) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/anchor_free/cornernet_squeeze_dcn_r50_vd_fpn.yml) |
| CornerNet-Squeeze-dcn-mixup-cosine* | ResNet50-vd | 14 | [faster\_rcnn\_dcn\_r50\_vd\_fpn\_2x](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_dcn_r50_vd_fpn_2x.tar) | 38.2 | 40.05 | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/cornernet_squeeze_dcn_r50_vd_fpn_mixup_cosine.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/anchor_free/cornernet_squeeze_dcn_r50_vd_fpn_mixup_cosine.yml) |
| FCOS | ResNet50 | 2 | [ResNet50\_cos\_pretrained](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar) | 39.8 | - | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/fcos_r50_fpn_1x.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/anchor_free/fcos_r50_fpn_1x.yml) |
| FCOS+multiscale_train | ResNet50 | 2 | [ResNet50\_cos\_pretrained](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar) | 42.0 | - | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/fcos_r50_fpn_multiscale_2x.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/anchor_free/fcos_r50_fpn_multiscale_2x.yml) |
| FCOS+DCN | ResNet50 | 2 | [ResNet50\_cos\_pretrained](https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar) | 44.4 | - | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/fcos_dcn_r50_fpn_1x.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/anchor_free/fcos_dcn_r50_fpn_1x.yml) |
**注意:**
- 模型FPS在Tesla V100单卡环境中通过tools/eval.py进行测试
- CornerNet-Squeeze要求使用PaddlePaddle1.8及以上版本或适当的develop版本
- CornerNet-Squeeze中使用ResNet结构的骨干网络时，加入了FPN结构，骨干网络的输出feature map采用FPN中的P3层输出。
- \*CornerNet-Squeeze-dcn-mixup-cosine是基于原版CornerNet-Squeeze优化效果最好的模型，在ResNet的骨干网络基础上增加mixup预处理和使用cosine_decay
- FCOS使用GIoU loss、用location分支预测centerness、左上右下角点偏移量归一化和ground truth中心匹配策略
- Cornernet-Squeeze模型依赖corner_pooling op，该op在```ppdet/ext_op```中编译得到，具体编译方式请参考[自定义OP的编译过程](../../ppdet/ext_op/README.md)
## 算法细节
### CornerNet-Squeeze
**简介:** [CornerNet-Squeeze](https://arxiv.org/abs/1904.08900) 在[Cornernet](https://arxiv.org/abs/1808.01244)基础上进行改进，预测目标框的左上角和右下角的位置，同时参考SqueezeNet和MobileNet的特点，优化了CornerNet骨干网络Hourglass-104，大幅提升了模型预测速度，相较于原版[YOLO-v3](https://arxiv.org/abs/1804.02767)，在训练精度和推理速度上都具备一定优势。
**特点:**
- 使用corner_pooling获取候选框左上角和右下角的位置
- 替换Hourglass-104中的residual block为SqueezeNet中的fire-module
- 替换第二层3x3卷积为3x3深度可分离卷积
### FCOS
**简介:** [FCOS](https://arxiv.org/abs/1904.01355)是一种密集预测的anchor-free检测算法，使用RetinaNet的骨架，直接在feature map上回归目标物体的长宽，并预测物体的类别以及centerness（feature map上像素点离物体中心的偏移程度），centerness最终会作为权重来调整物体得分。
**特点:**
- 利用FPN结构在不同层预测不同scale的物体框，避免了同一feature map像素点处有多个物体框重叠的情况
- 通过center-ness单层分支预测当前点是否是目标中心，消除低质量误检
## 如何贡献代码
我们非常欢迎您可以为PaddleDetection中的Anchor Free检测模型提供代码，您可以提交PR供我们review；也十分感谢您的反馈，可以提交相应issue，我们会及时解答。
configs/anchor_free/cornernet_squeeze_dcn_r50_vd_fpn.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CornerNetSqueeze
use_gpu: true
max_iters: 500000
log_smooth_window: 20
log_iter: 20
save_dir: output
snapshot_iter: 10000
metric: COCO
pretrain_weights: https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_dcn_r50_vd_fpn_2x.tar
weights: output/cornernet_squeeze_dcn_r50_vd_fpn/model_final
num_classes: 80
stack: 1
CornerNetSqueeze:
backbone: ResNet
fpn: FPN
corner_head: CornerHead
ResNet:
norm_type: bn
depth: 50
feature_maps: [3, 4, 5]
freeze_at: 2
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
min_level: 3
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125]
CornerHead:
train_batch_size: 14
test_batch_size: 1
ae_threshold: 0.5
num_dets: 100
top_k: 20
PostProcess:
use_soft_nms: true
detections_per_im: 100
nms_thresh: 0.001
sigma: 0.5
LearningRate:
base_lr: 0.0005
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones:
- 400000
- 450000
- !LinearWarmup
start_factor: 0.
steps: 4000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0005
type: L2
TrainReader:
inputs_def:
image_shape: [3, 511, 511]
fields: ['image', 'im_id', 'gt_bbox', 'gt_class', 'tl_heatmaps', 'br_heatmaps', 'tl_regrs', 'br_regrs', 'tl_tags', 'br_tags', 'tag_masks']
output_size: 64
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: False
- !CornerCrop
input_size: 511
- !Resize
target_dim: 511
- !RandomFlipImage
prob: 0.5
- !CornerRandColor
saturation: 0.4
contrast: 0.4
brightness: 0.4
- !Lighting
eigval: [0.2141788, 0.01817699, 0.00341571]
eigvec: [[-0.58752847, -0.69563484, 0.41340352],
[-0.5832747, 0.00994535, -0.81221408],
[-0.56089297, 0.71832671, 0.41158938]]
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: False
is_channel_first: False
- !Permute
to_bgr: False
- !CornerTarget
output_size: [64, 64]
num_classes: 80
batch_size: 14
shuffle: true
drop_last: true
worker_num: 2
use_process: true
drop_empty: false
EvalReader:
inputs_def:
fields: ['image', 'im_id', 'ratios', 'borders']
output_size: 64
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: false
- !CornerCrop
is_train: false
- !CornerRatio
input_size: 511
output_size: 64
- !Permute
to_bgr: False
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: True
is_channel_first: True
use_process: true
batch_size: 1
drop_empty: false
worker_num: 2
TestReader:
inputs_def:
fields: ['image', 'im_id', 'ratios', 'borders']
output_size: 64
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: false
- !CornerCrop
is_train: false
- !CornerRatio
input_size: 511
output_size: 64
- !Permute
to_bgr: False
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: True
is_channel_first: True
batch_size: 1
configs/anchor_free/cornernet_squeeze_dcn_r50_vd_fpn_mixup_cosine.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CornerNetSqueeze
use_gpu: true
max_iters: 500000
log_smooth_window: 20
log_iter: 20
save_dir: output
snapshot_iter: 10000
metric: COCO
pretrain_weights: https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_dcn_r50_vd_fpn_2x.tar
weights: output/cornernet_squeeze_dcn_r50_vd_fpn_mixup_cosine/model_final
num_classes: 80
stack: 1
CornerNetSqueeze:
backbone: ResNet
fpn: FPN
corner_head: CornerHead
ResNet:
norm_type: bn
depth: 50
feature_maps: [3, 4, 5]
freeze_at: 2
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
min_level: 3
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125]
CornerHead:
train_batch_size: 14
test_batch_size: 1
ae_threshold: 0.5
num_dets: 100
top_k: 20
PostProcess:
use_soft_nms: true
detections_per_im: 100
nms_thresh: 0.001
sigma: 0.5
LearningRate:
base_lr: 0.005
schedulers:
- !CosineDecay
max_iters: 500000
- !LinearWarmup
start_factor: 0.
steps: 4000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0005
type: L2
TrainReader:
inputs_def:
image_shape: [3, 511, 511]
fields: ['image', 'im_id', 'gt_bbox', 'gt_class', 'tl_heatmaps', 'br_heatmaps', 'tl_regrs', 'br_regrs', 'tl_tags', 'br_tags', 'tag_masks']
output_size: 64
max_tag_len: 256
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: False
with_mixup: True
- !MixupImage
alpha: 1.5
beta: 1.5
- !CornerCrop
input_size: 511
- !Resize
target_dim: 511
- !RandomFlipImage
prob: 0.5
- !CornerRandColor
saturation: 0.4
contrast: 0.4
brightness: 0.4
- !Lighting
eigval: [0.2141788, 0.01817699, 0.00341571]
eigvec: [[-0.58752847, -0.69563484, 0.41340352],
[-0.5832747, 0.00994535, -0.81221408],
[-0.56089297, 0.71832671, 0.41158938]]
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: False
is_channel_first: False
- !Permute
to_bgr: False
- !CornerTarget
output_size: [64, 64]
num_classes: 80
max_tag_len: 256
batch_size: 14
shuffle: true
drop_last: true
worker_num: 2
use_process: true
drop_empty: false
mixup_epoch: 200
EvalReader:
inputs_def:
fields: ['image', 'im_id', 'ratios', 'borders']
output_size: 64
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: false
- !CornerCrop
is_train: false
- !CornerRatio
input_size: 511
output_size: 64
- !Permute
to_bgr: False
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: True
is_channel_first: True
use_process: true
batch_size: 1
drop_empty: false
worker_num: 2
TestReader:
inputs_def:
fields: ['image', 'im_id', 'ratios', 'borders']
output_size: 64
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: false
- !CornerCrop
is_train: false
- !CornerRatio
input_size: 511
output_size: 64
- !Permute
to_bgr: False
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: True
is_channel_first: True
batch_size: 1
configs/anchor_free/cornernet_squeeze_hg104.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CornerNetSqueeze
use_gpu: true
max_iters: 500000
log_smooth_window: 20
log_iter: 20
save_dir: output
snapshot_iter: 10000
metric: COCO
pretrain_weights: NULL
weights: output/cornernet_squeeze_hg104/model_final
num_classes: 80
stack: 2
CornerNetSqueeze:
backbone: Hourglass
corner_head: CornerHead
Hourglass:
dims: [256, 256, 384, 384, 512]
modules: [2, 2, 2, 2, 4]
CornerHead:
train_batch_size: 14
test_batch_size: 1
ae_threshold: 0.5
num_dets: 100
top_k: 20
PostProcess:
use_soft_nms: true
detections_per_im: 100
nms_thresh: 0.001
sigma: 0.5
LearningRate:
base_lr: 0.00025
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones:
- 450000
OptimizerBuilder:
optimizer:
type: Adam
regularizer: NULL
TrainReader:
inputs_def:
image_shape: [3, 511, 511]
fields: ['image', 'im_id', 'gt_bbox', 'gt_class', 'tl_heatmaps', 'br_heatmaps', 'tl_regrs', 'br_regrs', 'tl_tags', 'br_tags', 'tag_masks']
output_size: 64
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: False
- !CornerCrop
input_size: 511
- !Resize
target_dim: 511
- !RandomFlipImage
prob: 0.5
- !CornerRandColor
saturation: 0.4
contrast: 0.4
brightness: 0.4
- !Lighting
eigval: [0.2141788, 0.01817699, 0.00341571]
eigvec: [[-0.58752847, -0.69563484, 0.41340352],
[-0.5832747, 0.00994535, -0.81221408],
[-0.56089297, 0.71832671, 0.41158938]]
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: False
is_channel_first: False
- !Permute
to_bgr: False
- !CornerTarget
output_size: [64, 64]
num_classes: 80
batch_size: 14
shuffle: true
drop_last: true
worker_num: 2
use_process: true
drop_empty: false
EvalReader:
inputs_def:
fields: ['image', 'im_id', 'ratios', 'borders']
output_size: 64
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: false
- !CornerCrop
is_train: false
- !CornerRatio
input_size: 511
output_size: 64
- !Permute
to_bgr: False
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: True
is_channel_first: True
batch_size: 1
drop_empty: false
worker_num: 2
use_process: true
TestReader:
inputs_def:
fields: ['image', 'im_id', 'ratios', 'borders']
output_size: 64
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: false
- !CornerCrop
is_train: false
- !CornerRatio
input_size: 511
output_size: 64
- !Permute
to_bgr: False
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: True
is_channel_first: True
batch_size: 1
configs/anchor_free/cornernet_squeeze_r50_vd_fpn.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CornerNetSqueeze
use_gpu: true
max_iters: 500000
log_smooth_window: 20
log_iter: 20
save_dir: output
snapshot_iter: 10000
metric: COCO
pretrain_weights: https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_dcn_r50_vd_fpn_2x.tar
weights: output/cornernet_squeeze_r50_vd_fpn/model_final
num_classes: 80
stack: 1
CornerNetSqueeze:
backbone: ResNet
fpn: FPN
corner_head: CornerHead
ResNet:
norm_type: affine_channel
depth: 50
feature_maps: [3, 4, 5]
freeze_at: 2
variant: d
FPN:
min_level: 3
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125]
CornerHead:
train_batch_size: 14
test_batch_size: 1
ae_threshold: 0.5
num_dets: 100
top_k: 20
PostProcess:
use_soft_nms: true
detections_per_im: 100
nms_thresh: 0.001
sigma: 0.5
LearningRate:
base_lr: 0.0005
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones:
- 450000
OptimizerBuilder:
optimizer:
type: Adam
regularizer: NULL
TrainReader:
inputs_def:
image_shape: [3, 511, 511]
fields: ['image', 'im_id', 'gt_bbox', 'gt_class', 'tl_heatmaps', 'br_heatmaps', 'tl_regrs', 'br_regrs', 'tl_tags', 'br_tags', 'tag_masks']
output_size: 64
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: False
- !CornerCrop
input_size: 511
- !Resize
target_dim: 511
- !RandomFlipImage
prob: 0.5
- !CornerRandColor
saturation: 0.4
contrast: 0.4
brightness: 0.4
- !Lighting
eigval: [0.2141788, 0.01817699, 0.00341571]
eigvec: [[-0.58752847, -0.69563484, 0.41340352],
[-0.5832747, 0.00994535, -0.81221408],
[-0.56089297, 0.71832671, 0.41158938]]
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: False
is_channel_first: False
- !Permute
to_bgr: False
- !CornerTarget
output_size: [64, 64]
num_classes: 80
batch_size: 14
shuffle: true
drop_last: true
worker_num: 2
use_process: true
drop_empty: false
EvalReader:
inputs_def:
fields: ['image', 'im_id', 'ratios', 'borders']
output_size: 64
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: false
- !CornerCrop
is_train: false
- !CornerRatio
input_size: 511
output_size: 64
- !Permute
to_bgr: False
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: True
is_channel_first: True
use_process: true
batch_size: 1
drop_empty: false
worker_num: 2
TestReader:
inputs_def:
fields: ['image', 'im_id', 'ratios', 'borders']
output_size: 64
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: false
- !CornerCrop
is_train: false
- !CornerRatio
input_size: 511
output_size: 64
- !Permute
to_bgr: False
- !NormalizeImage
mean: [0.40789654, 0.44719302, 0.47026115]
std: [0.28863828, 0.27408164, 0.2780983]
is_scale: True
is_channel_first: True
batch_size: 1
configs/anchor_free/fcos_dcn_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: FCOS
max_iters: 90000
use_gpu: true
snapshot_iter: 5000
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
metric: COCO
weights: output/fcos_dcn_r50_fpn_1x/model_final
num_classes: 80
FCOS:
backbone: ResNet
fpn: FPN
fcos_head: FCOSHead
ResNet:
norm_type: affine_channel
norm_decay: 0.
depth: 50
feature_maps: [3, 4, 5]
freeze_at: 2
dcn_v2_stages: [3, 4, 5]
FPN:
min_level: 3
max_level: 7
num_chan: 256
use_c5: false
spatial_scale: [0.03125, 0.0625, 0.125]
has_extra_convs: true
FCOSHead:
num_classes: 80
fpn_stride: [8, 16, 32, 64, 128]
num_convs: 4
norm_type: "gn"
fcos_loss: FCOSLoss
norm_reg_targets: True
centerness_on_reg: True
use_dcn_in_tower: True
nms: MultiClassNMS
MultiClassNMS:
score_threshold: 0.025
nms_top_k: 1000
keep_top_k: 100
nms_threshold: 0.6
background_label: -1
FCOSLoss:
loss_alpha: 0.25
loss_gamma: 2.0
iou_loss_type: "giou"
reg_weights: 1.0
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.3333333333333333
steps: 500
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
TrainReader:
inputs_def:
fields: ['image', 'gt_bbox', 'gt_class', 'gt_score', 'im_info']
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: true
- !RandomFlipImage
prob: 0.5
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
target_size: 800
max_size: 1333
interp: 1
use_cv2: true
- !Permute
to_bgr: false
channel_first: true
batch_transforms:
- !PadBatch
pad_to_stride: 128
use_padded_im_info: false
- !Gt2FCOSTarget
object_sizes_boundary: [64, 128, 256, 512]
center_sampling_radius: 1.5
downsample_ratios: [8, 16, 32, 64, 128]
norm_reg_targets: True
batch_size: 2
shuffle: true
worker_num: 4
use_process: false
EvalReader:
inputs_def:
fields: ['image', 'im_id', 'im_shape', 'im_info']
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
target_size: 800
max_size: 1333
interp: 1
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 128
use_padded_im_info: true
batch_size: 1
shuffle: false
worker_num: 1
use_process: false
TestReader:
inputs_def:
# set image_shape if needed
fields: ['image', 'im_id', 'im_shape', 'im_info']
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
interp: 1
max_size: 1333
target_size: 800
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 128
use_padded_im_info: true
batch_size: 1
shuffle: false
configs/anchor_free/fcos_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: FCOS
max_iters: 90000
use_gpu: true
snapshot_iter: 10000
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
metric: COCO
weights: output/fcos_r50_fpn_1x/model_final
num_classes: 80
FCOS:
backbone: ResNet
fpn: FPN
fcos_head: FCOSHead
ResNet:
norm_type: affine_channel
norm_decay: 0.
depth: 50
feature_maps: [3, 4, 5]
freeze_at: 2
FPN:
min_level: 3
max_level: 7
num_chan: 256
use_c5: false
spatial_scale: [0.03125, 0.0625, 0.125]
has_extra_convs: true
FCOSHead:
num_classes: 80
fpn_stride: [8, 16, 32, 64, 128]
num_convs: 4
norm_type: "gn"
fcos_loss: FCOSLoss
norm_reg_targets: True
centerness_on_reg: True
use_dcn_in_tower: False
nms: MultiClassNMS
MultiClassNMS:
score_threshold: 0.025
nms_top_k: 1000
keep_top_k: 100
nms_threshold: 0.6
background_label: -1
FCOSLoss:
loss_alpha: 0.25
loss_gamma: 2.0
iou_loss_type: "giou"
reg_weights: 1.0
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.3333333333333333
steps: 500
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
TrainReader:
inputs_def:
fields: ['image', 'gt_bbox', 'gt_class', 'gt_score', 'im_info']
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: true
- !RandomFlipImage
prob: 0.5
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
target_size: 800
max_size: 1333
interp: 1
use_cv2: true
- !Permute
to_bgr: false
channel_first: true
batch_transforms:
- !PadBatch
pad_to_stride: 128
use_padded_im_info: false
- !Gt2FCOSTarget
object_sizes_boundary: [64, 128, 256, 512]
center_sampling_radius: 1.5
downsample_ratios: [8, 16, 32, 64, 128]
norm_reg_targets: True
batch_size: 2
shuffle: true
worker_num: 4
use_process: false
EvalReader:
inputs_def:
fields: ['image', 'im_id', 'im_shape', 'im_info']
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
target_size: 800
max_size: 1333
interp: 1
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 128
use_padded_im_info: true
batch_size: 1
shuffle: false
worker_num: 2
use_process: false
TestReader:
inputs_def:
# set image_shape if needed
fields: ['image', 'im_id', 'im_shape', 'im_info']
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
interp: 1
max_size: 1333
target_size: 800
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 128
use_padded_im_info: true
batch_size: 1
shuffle: false
configs/anchor_free/fcos_r50_fpn_multiscale_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: FCOS
max_iters: 180000
use_gpu: true
snapshot_iter: 20000
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
metric: COCO
weights: output/fcos_r50_fpn_multiscale_2x/model_final
num_classes: 80
FCOS:
backbone: ResNet
fpn: FPN
fcos_head: FCOSHead
ResNet:
norm_type: affine_channel
norm_decay: 0.
depth: 50
feature_maps: [3, 4, 5]
freeze_at: 2
FPN:
min_level: 3
max_level: 7
num_chan: 256
use_c5: false
spatial_scale: [0.03125, 0.0625, 0.125]
has_extra_convs: true
FCOSHead:
num_classes: 80
fpn_stride: [8, 16, 32, 64, 128]
num_convs: 4
norm_type: "gn"
fcos_loss: FCOSLoss
norm_reg_targets: True
centerness_on_reg: True
use_dcn_in_tower: False
nms: MultiClassNMS
MultiClassNMS:
score_threshold: 0.025
nms_top_k: 1000
keep_top_k: 100
nms_threshold: 0.6
background_label: -1
FCOSLoss:
loss_alpha: 0.25
loss_gamma: 2.0
iou_loss_type: "giou"
reg_weights: 1.0
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [120000, 160000]
- !LinearWarmup
start_factor: 0.3333333333333333
steps: 500
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
TrainReader:
inputs_def:
fields: ['image', 'gt_bbox', 'gt_class', 'gt_score', 'im_info']
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: true
- !RandomFlipImage
prob: 0.5
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
target_size: [640, 672, 704, 736, 768, 800]
max_size: 1333
interp: 1
use_cv2: true
- !Permute
to_bgr: false
channel_first: true
batch_transforms:
- !PadBatch
pad_to_stride: 128
use_padded_im_info: false
- !Gt2FCOSTarget
object_sizes_boundary: [64, 128, 256, 512]
center_sampling_radius: 1.5
downsample_ratios: [8, 16, 32, 64, 128]
norm_reg_targets: True
batch_size: 2
shuffle: true
worker_num: 4
use_process: false
EvalReader:
inputs_def:
fields: ['image', 'im_id', 'im_shape', 'im_info']
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
target_size: 800
max_size: 1333
interp: 1
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 128
use_padded_im_info: true
batch_size: 1
shuffle: false
worker_num: 2
use_process: false
TestReader:
inputs_def:
# set image_shape if needed
fields: ['image', 'im_id', 'im_shape', 'im_info']
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
interp: 1
max_size: 1333
target_size: 800
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 128
use_padded_im_info: true
batch_size: 1
shuffle: false
configs/autoaugment/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
# Learning Data Augmentation Strategies for Object Detection
## Introduction
- Learning Data Augmentation Strategies for Object Detection: [https://arxiv.org/abs/1906.11172](https://arxiv.org/abs/1906.11172)
```
@article{Zoph2019LearningDA,
title={Learning Data Augmentation Strategies for Object Detection},
author={Barret Zoph and Ekin Dogus Cubuk and Golnaz Ghiasi and Tsung-Yi Lin and Jonathon Shlens and Quoc V. Le},
journal={ArXiv},
year={2019},
volume={abs/1906.11172}
}
```
## Model Zoo
| Backbone | Type | AutoAug policy | Image/gpu | Lr schd | Inf time (fps) | Box AP | Mask AP | Download | Configs |
| :---------------------- | :-------------:| :-------: | :-------: | :-----: | :------------: | :----: | :-----: | :----------------------------------------------------------: | :-----: |
| ResNet50-vd-FPN | Faster | v1 | 2 | 3x | 22.800 | 39.9 | - | [model](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r50_vd_fpn_aa_3x.tar) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/autoaugment/faster_rcnn_r50_vd_fpn_aa_3x.yml) |
| ResNet101-vd-FPN | Faster | v1 | 2 | 3x | 17.652 | 42.5 | - | [model](https://paddlemodels.bj.bcebos.com/object_detection/faster_rcnn_r101_vd_fpn_aa_3x.tar) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/autoaugment/faster_rcnn_r101_vd_fpn_aa_3x.yml) |
configs/autoaugment/faster_rcnn_r101_vd_fpn_aa_3x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: FasterRCNN
max_iters: 270000
snapshot_iter: 30000
use_gpu: true
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_vd_pretrained.tar
weights: output/faster_rcnn_r101_vd_fpn_aa_3x/model_final
metric: COCO
num_classes: 81
FasterRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNet:
depth: 101
feature_maps: [2, 3, 4, 5]
freeze_at: 2
norm_type: bn
variant: d
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 2000
pre_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 1000
pre_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
box_resolution: 7
sampling_ratio: 2
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_hi: 0.5
bg_thresh_lo: 0.0
fg_fraction: 0.25
fg_thresh: 0.5
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [180000, 240000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../faster_fpn_reader.yml'
TrainReader:
sample_transforms:
- !DecodeImage
to_rgb: true
- !RandomFlipImage
prob: 0.5
- !AutoAugmentImage
autoaug_type: v1
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
target_size: 800
max_size: 1333
interp: 1
use_cv2: true
- !Permute
to_bgr: false
channel_first: true
batch_size: 2
use_process: true
configs/autoaugment/faster_rcnn_r50_vd_fpn_aa_3x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: FasterRCNN
max_iters: 270000
snapshot_iter: 30000
use_gpu: true
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_pretrained.tar
weights: output/faster_rcnn_r50_vd_fpn_aa_3x/model_final
metric: COCO
num_classes: 81
FasterRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNet:
depth: 50
feature_maps: [2, 3, 4, 5]
freeze_at: 2
norm_type: bn
variant: d
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 2000
pre_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 1000
pre_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
box_resolution: 7
sampling_ratio: 2
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_hi: 0.5
bg_thresh_lo: 0.0
fg_fraction: 0.25
fg_thresh: 0.5
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [180000, 240000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../faster_fpn_reader.yml'
TrainReader:
sample_transforms:
- !DecodeImage
to_rgb: true
- !RandomFlipImage
prob: 0.5
- !AutoAugmentImage
autoaug_type: v1
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
target_size: 800
max_size: 1333
interp: 1
use_cv2: true
- !Permute
to_bgr: false
channel_first: true
batch_size: 2
use_process: true
configs/cascade_mask_rcnn_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeMaskRCNN
use_gpu: true
max_iters: 180000
snapshot_iter: 10000
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
metric: COCO
weights: output/cascade_mask_rcnn_r50_fpn_1x/model_final
num_classes: 81
CascadeMaskRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
mask_assigner: MaskAssigner
mask_head: MaskHead
ResNet:
depth: 50
feature_maps: [2, 3, 4, 5]
freeze_at: 2
norm_type: affine_channel
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
aspect_ratios: [0.5, 1.0, 2.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
sampling_ratio: 2
box_resolution: 7
mask_resolution: 14
MaskHead:
dilation: 1
conv_dim: 256
num_convs: 4
resolution: 28
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_hi: [0.5, 0.6, 0.7]
bg_thresh_lo: [0.0, 0.0, 0.0]
fg_fraction: 0.25
fg_thresh: [0.5, 0.6, 0.7]
MaskAssigner:
resolution: 28
CascadeBBoxHead:
head: CascadeTwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
CascadeTwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [120000, 160000]
- !LinearWarmup
start_factor: 0.3333333333333333
steps: 500
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: 'mask_fpn_reader.yml'
configs/cascade_rcnn_cls_aware_r101_vd_fpn_1x_softnms.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeRCNNClsAware
max_iters: 90000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_vd_pretrained.tar
weights: output/cascade_rcnn_cls_aware_r101_vd_fpn_1x_softnms/model_final
metric: COCO
num_classes: 81
CascadeRCNNClsAware:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
ResNet:
norm_type: bn
depth: 101
feature_maps: [2, 3, 4, 5]
freeze_at: 2
variant: d
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 14
sampling_ratio: 2
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_lo: [0.0, 0.0, 0.0]
bg_thresh_hi: [0.5, 0.6, 0.7]
fg_thresh: [0.5, 0.6, 0.7]
fg_fraction: 0.25
class_aware: True
CascadeBBoxHead:
head: CascadeTwoFCHead
nms: MultiClassSoftNMS
CascadeTwoFCHead:
mlp_dim: 1024
MultiClassSoftNMS:
score_threshold: 0.01
keep_top_k: 300
softnms_sigma: 0.5
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.0
steps: 2000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: 'faster_fpn_reader.yml'
TrainReader:
batch_size: 2
configs/cascade_rcnn_cls_aware_r101_vd_fpn_ms_test.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeRCNNClsAware
max_iters: 90000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_vd_pretrained.tar
weights: output/cascade_rcnn_cls_aware_r101_vd_fpn_ms_test/model_final
metric: COCO
num_classes: 81
CascadeRCNNClsAware:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
ResNet:
norm_type: bn
depth: 101
feature_maps: [2, 3, 4, 5]
freeze_at: 2
variant: d
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 14
sampling_ratio: 2
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_lo: [0.0, 0.0, 0.0]
bg_thresh_hi: [0.5, 0.6, 0.7]
fg_thresh: [0.5, 0.6, 0.7]
fg_fraction: 0.25
class_aware: True
CascadeBBoxHead:
head: CascadeTwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
CascadeTwoFCHead:
mlp_dim: 1024
MultiScaleTEST:
score_thresh: 0.05
nms_thresh: 0.5
detections_per_im: 100
enable_voting: true
vote_thresh: 0.9
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.0
steps: 2000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
EvalReader:
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
multi_scale: true
num_scales: 18
use_flip: true
dataset:
!COCODataSet
dataset_dir: dataset/coco
anno_path: annotations/instances_val2017.json
image_dir: val2017
sample_transforms:
- !DecodeImage
to_rgb: true
- !NormalizeImage
is_channel_first: false
is_scale: true
mean:
- 0.485
- 0.456
- 0.406
std:
- 0.229
- 0.224
- 0.225
- !MultiscaleTestResize
origin_target_size: 800
origin_max_size: 1333
target_size:
- 400
- 500
- 600
- 700
- 900
- 1000
- 1100
- 1200
max_size: 2000
use_flip: true
- !Permute
channel_first: true
to_bgr: false
- !PadMultiScaleTest
pad_to_stride: 32
worker_num: 2
configs/cascade_rcnn_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeRCNN
max_iters: 90000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
weights: output/cascade_rcnn_r50_fpn_1x/model_final
metric: COCO
num_classes: 81
CascadeRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
ResNet:
norm_type: affine_channel
depth: 50
feature_maps: [2, 3, 4, 5]
freeze_at: 2
variant: b
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 7
sampling_ratio: 2
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_lo: [0.0, 0.0, 0.0]
bg_thresh_hi: [0.5, 0.6, 0.7]
fg_thresh: [0.5, 0.6, 0.7]
fg_fraction: 0.25
CascadeBBoxHead:
head: CascadeTwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
CascadeTwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.3333333333333333
steps: 500
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: 'faster_fpn_reader.yml'
TrainReader:
batch_size: 2
configs/cascade_rcnn_r50_fpn_1x_ms_test.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeRCNN
max_iters: 90000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
weights: output/cascade_rcnn_r50_fpn_1x/model_final
metric: COCO
num_classes: 81
CascadeRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
ResNet:
norm_type: affine_channel
depth: 50
feature_maps: [2, 3, 4, 5]
freeze_at: 2
variant: b
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 7
sampling_ratio: 2
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_lo: [0.0, 0.0, 0.0]
bg_thresh_hi: [0.5, 0.6, 0.7]
fg_thresh: [0.5, 0.6, 0.7]
fg_fraction: 0.25
CascadeBBoxHead:
head: CascadeTwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
CascadeTwoFCHead:
mlp_dim: 1024
MultiScaleTEST:
score_thresh: 0.05
nms_thresh: 0.5
detections_per_im: 100
enable_voting: true
vote_thresh: 0.9
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.3333333333333333
steps: 500
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: 'faster_fpn_reader.yml'
TrainReader:
batch_size: 2
EvalReader:
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
multi_scale: true
num_scales: 18
use_flip: true
dataset:
!COCODataSet
dataset_dir: dataset/coco
anno_path: annotations/instances_val2017.json
image_dir: val2017
sample_transforms:
- !DecodeImage
to_rgb: true
- !NormalizeImage
is_channel_first: false
is_scale: true
mean:
- 0.485
- 0.456
- 0.406
std:
- 0.229
- 0.224
- 0.225
- !MultiscaleTestResize
origin_target_size: 800
origin_max_size: 1333
target_size:
- 400
- 500
- 600
- 700
- 900
- 1000
- 1100
- 1200
max_size: 2000
use_flip: true
- !Permute
channel_first: true
to_bgr: false
- !PadMultiScaleTest
pad_to_stride: 32
worker_num: 2
configs/dcn/cascade_mask_rcnn_dcnv2_se154_vd_fpn_gn_s1x.yml 已删除 100755 → 0
浏览文件 @ 7cf31190
architecture: CascadeMaskRCNN
max_iters: 300000
snapshot_iter: 10
use_gpu: true
log_iter: 20
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/SENet154_vd_caffe_pretrained.tar
weights: output/cascade_mask_rcnn_dcn_se154_vd_fpn_gn_s1x/model_final
metric: COCO
num_classes: 81
CascadeMaskRCNN:
backbone: SENet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
mask_assigner: MaskAssigner
mask_head: MaskHead
SENet:
depth: 152
feature_maps: [2, 3, 4, 5]
freeze_at: 2
group_width: 4
groups: 64
norm_type: bn
freeze_norm: True
variant: d
dcn_v2_stages: [3, 4, 5]
std_senet: True
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
freeze_norm: False
norm_type: gn
FPNRPNHead:
anchor_generator:
aspect_ratios: [0.5, 1.0, 2.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
box_resolution: 7
sampling_ratio: 2
mask_resolution: 14
MaskHead:
dilation: 1
conv_dim: 256
num_convs: 4
resolution: 28
norm_type: gn
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_hi: [0.5, 0.6, 0.7]
bg_thresh_lo: [0.0, 0.0, 0.0]
fg_fraction: 0.25
fg_thresh: [0.5, 0.6, 0.7]
MaskAssigner:
resolution: 28
CascadeBBoxHead:
head: CascadeXConvNormHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
CascadeXConvNormHead:
norm_type: gn
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [240000, 280000]
- !LinearWarmup
start_factor: 0.01
steps: 2000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
TrainReader:
# batch size per device
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'gt_bbox', 'gt_class', 'is_crowd', 'gt_mask']
dataset:
!COCODataSet
dataset_dir: dataset/coco
image_dir: train2017
anno_path: annotations/instances_train2017.json
sample_transforms:
- !DecodeImage
to_rgb: false
- !RandomFlipImage
is_mask_flip: true
is_normalized: false
prob: 0.5
- !NormalizeImage
is_channel_first: false
is_scale: False
mean:
- 102.9801
- 115.9465
- 122.7717
std:
- 1.0
- 1.0
- 1.0
- !ResizeImage
interp: 1
target_size:
- 416
- 448
- 480
- 512
- 544
- 576
- 608
- 640
- 672
- 704
- 736
- 768
- 800
- 832
- 864
- 896
- 928
- 960
- 992
- 1024
- 1056
- 1088
- 1120
- 1152
- 1184
- 1216
- 1248
- 1280
- 1312
- 1344
- 1376
- 1408
max_size: 1600
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
worker_num: 8
shuffle: true
EvalReader:
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
dataset:
!COCODataSet
dataset_dir: dataset/coco
anno_path: annotations/instances_val2017.json
image_dir: val2017
sample_transforms:
- !DecodeImage
to_rgb: False
- !NormalizeImage
is_channel_first: false
is_scale: False
mean:
- 102.9801
- 115.9465
- 122.7717
std:
- 1.0
- 1.0
- 1.0
- !ResizeImage
interp: 1
target_size:
- 800
max_size: 1333
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
worker_num: 2
drop_empty: false
TestReader:
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
sample_transforms:
- !DecodeImage
to_rgb: False
- !NormalizeImage
is_channel_first: false
is_scale: False
mean:
- 102.9801
- 115.9465
- 122.7717
std:
- 1.0
- 1.0
- 1.0
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
worker_num: 2
configs/dcn/cascade_mask_rcnn_dcnv2_se154_vd_fpn_gn_s1x_ms_test.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeMaskRCNN
max_iters: 300000
snapshot_iter: 10000
use_gpu: true
log_iter: 20
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/SENet154_vd_caffe_pretrained.tar
weights: output/cascade_mask_rcnn_dcn_se154_vd_fpn_gn_s1x/model_final
metric: COCO
num_classes: 81
CascadeMaskRCNN:
backbone: SENet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
mask_assigner: MaskAssigner
mask_head: MaskHead
SENet:
depth: 152
feature_maps: [2, 3, 4, 5]
freeze_at: 2
group_width: 4
groups: 64
norm_type: bn
freeze_norm: True
variant: d
dcn_v2_stages: [3, 4, 5]
std_senet: True
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
freeze_norm: False
norm_type: gn
FPNRPNHead:
anchor_generator:
aspect_ratios: [0.5, 1.0, 2.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
box_resolution: 7
sampling_ratio: 2
mask_resolution: 14
MaskHead:
dilation: 1
conv_dim: 256
num_convs: 4
resolution: 28
norm_type: gn
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_hi: [0.5, 0.6, 0.7]
bg_thresh_lo: [0.0, 0.0, 0.0]
fg_fraction: 0.25
fg_thresh: [0.5, 0.6, 0.7]
MaskAssigner:
resolution: 28
CascadeBBoxHead:
head: CascadeXConvNormHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
CascadeXConvNormHead:
norm_type: gn
MultiScaleTEST:
score_thresh: 0.05
nms_thresh: 0.5
detections_per_im: 100
enable_voting: true
vote_thresh: 0.9
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [240000, 280000]
- !LinearWarmup
start_factor: 0.01
steps: 2000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
TrainReader:
# batch size per device
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'gt_bbox', 'gt_class', 'is_crowd', 'gt_mask']
dataset:
!COCODataSet
dataset_dir: dataset/coco
image_dir: train2017
anno_path: annotations/instances_train2017.json
sample_transforms:
- !DecodeImage
to_rgb: False
with_mixup: False
- !RandomFlipImage
is_mask_flip: true
is_normalized: false
prob: 0.5
- !NormalizeImage
is_channel_first: false
is_scale: False
mean:
- 102.9801
- 115.9465
- 122.7717
std:
- 1.0
- 1.0
- 1.0
- !ResizeImage
interp: 1
target_size:
- 416
- 448
- 480
- 512
- 544
- 576
- 608
- 640
- 672
- 704
- 736
- 768
- 800
- 832
- 864
- 896
- 928
- 960
- 992
- 1024
- 1056
- 1088
- 1120
- 1152
- 1184
- 1216
- 1248
- 1280
- 1312
- 1344
- 1376
- 1408
max_size: 1600
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
worker_num: 8
shuffle: true
EvalReader:
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
multi_scale: true
# num_scale = (len(target_size) + 1) * (1 + use_flip)
num_scales: 18
use_flip: true
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
sample_transforms:
- !DecodeImage
to_rgb: False
- !NormalizeImage
is_channel_first: false
is_scale: False
mean:
- 102.9801
- 115.9465
- 122.7717
std:
- 1.0
- 1.0
- 1.0
- !MultiscaleTestResize
origin_target_size: 800
origin_max_size: 1333
target_size:
- 400
- 500
- 600
- 700
- 900
- 1000
- 1100
- 1200
max_size: 2000
use_flip: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadMultiScaleTest
pad_to_stride: 32
worker_num: 2
TestReader:
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
sample_transforms:
- !DecodeImage
to_rgb: False
- !NormalizeImage
is_channel_first: false
is_scale: False
mean:
- 102.9801
- 115.9465
- 122.7717
std:
- 1.0
- 1.0
- 1.0
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
configs/dcn/cascade_rcnn_cbr200_vd_fpn_dcnv2_nonlocal_softnms.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeRCNN
max_iters: 460000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/CBResNet200_vd_pretrained.tar
weights: output/cascade_rcnn_cbr200_vd_fpn_dcnv2_nonlocal_softnms/model_final
metric: COCO
num_classes: 81
CascadeRCNN:
backbone: CBResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
CBResNet:
norm_type: bn
depth: 200
feature_maps: [2, 3, 4, 5]
freeze_at: 2
variant: d
dcn_v2_stages: [3, 4, 5]
nonlocal_stages: [4]
repeat_num: 2
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 14
sampling_ratio: 2
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_lo: [0.0, 0.0, 0.0]
bg_thresh_hi: [0.5, 0.6, 0.7]
fg_thresh: [0.5, 0.6, 0.7]
fg_fraction: 0.25
CascadeBBoxHead:
head: CascadeTwoFCHead
nms: MultiClassSoftNMS
CascadeTwoFCHead:
mlp_dim: 1024
MultiClassSoftNMS:
score_threshold: 0.01
keep_top_k: 300
softnms_sigma: 0.5
LearningRate:
base_lr: 0.005
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [340000, 440000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
TrainReader:
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'gt_bbox', 'gt_class', 'is_crowd']
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
sample_transforms:
- !DecodeImage
to_rgb: true
- !RandomFlipImage
prob: 0.5
- !NormalizeImage
is_channel_first: false
is_scale: True
mean:
- 0.485
- 0.456
- 0.406
std:
- 0.229
- 0.224
- 0.225
- !ResizeImage
interp: 1
target_size: [416, 448, 480, 512, 544, 576, 608, 640, 672, 704, 736, 768, 800, 832, 864, 896, 928, 960, 992, 1024, 1056, 1088, 1120, 1152, 1184, 1216, 1248, 1280, 1312, 1344, 1376, 1408]
max_size: 1600
use_cv2: true
- !Permute
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
worker_num: 2
shuffle: true
EvalReader:
batch_size: 1
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
sample_transforms:
- !DecodeImage
to_rgb: True
with_mixup: False
- !NormalizeImage
is_channel_first: false
is_scale: True
mean:
- 0.485
- 0.456
- 0.406
std:
- 0.229
- 0.224
- 0.225
- !ResizeImage
interp: 1
target_size:
- 1200
max_size: 2000
use_cv2: true
- !Permute
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
worker_num: 2
TestReader:
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
interp: 1
max_size: 1333
target_size: 800
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
use_padded_im_info: true
batch_size: 1
worker_num: 2
configs/dcn/cascade_rcnn_cls_aware_r200_vd_fpn_dcnv2_nonlocal_softnms.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeRCNNClsAware
max_iters: 460000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet200_vd_pretrained.tar
weights: output/cascade_rcnn_cls_aware_r200_vd_fpn_dcnv2_nonlocal_softnms/model_final
metric: COCO
num_classes: 81
CascadeRCNNClsAware:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
ResNet:
norm_type: bn
depth: 200
feature_maps: [2, 3, 4, 5]
freeze_at: 2
variant: d
dcn_v2_stages: [3, 4, 5]
nonlocal_stages: [4]
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 14
sampling_ratio: 2
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_lo: [0.0, 0.0, 0.0]
bg_thresh_hi: [0.5, 0.6, 0.7]
fg_thresh: [0.5, 0.6, 0.7]
fg_fraction: 0.25
class_aware: True
CascadeBBoxHead:
head: CascadeTwoFCHead
nms: MultiClassSoftNMS
CascadeTwoFCHead:
mlp_dim: 1024
MultiClassSoftNMS:
score_threshold: 0.01
keep_top_k: 300
softnms_sigma: 0.5
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [340000, 440000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
TrainReader:
inputs_def:
fields: ['image', 'im_info', 'im_id', 'gt_bbox', 'gt_class', 'is_crowd']
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
sample_transforms:
- !DecodeImage
to_rgb: true
- !RandomFlipImage
prob: 0.5
- !NormalizeImage
is_channel_first: false
is_scale: True
mean:
- 0.485
- 0.456
- 0.406
std:
- 0.229
- 0.224
- 0.225
- !ResizeImage
interp: 1
target_size: [416, 448, 480, 512, 544, 576, 608, 640, 672, 704, 736, 768, 800, 832, 864, 896, 928, 960, 992, 1024, 1056, 1088, 1120, 1152, 1184, 1216, 1248, 1280, 1312, 1344, 1376, 1408]
max_size: 1800
use_cv2: true
- !Permute
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
batch_size: 1
shuffle: true
drop_last: false
worker_num: 2
EvalReader:
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
sample_transforms:
- !DecodeImage
to_rgb: True
with_mixup: False
- !NormalizeImage
is_channel_first: false
is_scale: True
mean:
- 0.485
- 0.456
- 0.406
std:
- 0.229
- 0.224
- 0.225
- !ResizeImage
interp: 1
target_size:
- 1200
max_size: 2000
use_cv2: true
- !Permute
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
batch_size: 1
worker_num: 2
drop_empty: false
TestReader:
inputs_def:
fields: ['image', 'im_info', 'im_id', 'im_shape']
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeImage
interp: 1
max_size: 1333
target_size: 800
use_cv2: true
- !Permute
channel_first: true
to_bgr: false
batch_transforms:
- !PadBatch
pad_to_stride: 32
use_padded_im_info: true
batch_size: 1
worker_num: 2
configs/dcn/cascade_rcnn_dcn_r101_vd_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeRCNN
max_iters: 90000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_vd_pretrained.tar
weights: output/cascade_rcnn_dcn_r101_vd_fpn_1x/model_final
metric: COCO
num_classes: 81
CascadeRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
ResNet:
norm_type: bn
depth: 101
feature_maps: [2, 3, 4, 5]
freeze_at: 2
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 7
sampling_ratio: 2
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_lo: [0.0, 0.0, 0.0]
bg_thresh_hi: [0.5, 0.6, 0.7]
fg_thresh: [0.5, 0.6, 0.7]
fg_fraction: 0.25
CascadeBBoxHead:
head: CascadeTwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
CascadeTwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../faster_fpn_reader.yml'
TrainReader:
batch_size: 2
configs/dcn/cascade_rcnn_dcn_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeRCNN
max_iters: 90000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
weights: output/cascade_rcnn_dcn_r50_fpn_1x/model_final
metric: COCO
num_classes: 81
CascadeRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
ResNet:
norm_type: bn
depth: 50
feature_maps: [2, 3, 4, 5]
freeze_at: 2
variant: b
dcn_v2_stages: [3, 4, 5]
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 7
sampling_ratio: 2
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_lo: [0.0, 0.0, 0.0]
bg_thresh_hi: [0.5, 0.6, 0.7]
fg_thresh: [0.5, 0.6, 0.7]
fg_fraction: 0.25
CascadeBBoxHead:
head: CascadeTwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
CascadeTwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../faster_fpn_reader.yml'
TrainReader:
batch_size: 2
configs/dcn/cascade_rcnn_dcn_x101_vd_64x4d_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: CascadeRCNN
max_iters: 90000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNeXt101_vd_64x4d_pretrained.tar
weights: output/cascade_rcnn_dcn_x101_vd_64x4d_fpn_1x/model_final
metric: COCO
num_classes: 81
CascadeRCNN:
backbone: ResNeXt
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: CascadeBBoxHead
bbox_assigner: CascadeBBoxAssigner
ResNeXt:
norm_type: bn
depth: 101
feature_maps: [2, 3, 4, 5]
freeze_at: 2
group_width: 4
groups: 64
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 7
sampling_ratio: 2
CascadeBBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [10, 20, 30]
bg_thresh_lo: [0.0, 0.0, 0.0]
bg_thresh_hi: [0.5, 0.6, 0.7]
fg_thresh: [0.5, 0.6, 0.7]
fg_fraction: 0.25
CascadeBBoxHead:
head: CascadeTwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
CascadeTwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../faster_fpn_reader.yml'
TrainReader:
batch_size: 2
configs/dcn/faster_rcnn_dcn_r101_vd_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: FasterRCNN
max_iters: 90000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_vd_pretrained.tar
weights: output/faster_rcnn_dcn_r101_vd_fpn_1x/model_final
metric: COCO
num_classes: 81
FasterRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNet:
depth: 101
feature_maps: [2, 3, 4, 5]
freeze_at: 2
norm_type: bn
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 2000
pre_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 1000
pre_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
box_resolution: 7
sampling_ratio: 2
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_hi: 0.5
bg_thresh_lo: 0.0
fg_fraction: 0.25
fg_thresh: 0.5
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../faster_fpn_reader.yml'
TrainReader:
# batch size per device
batch_size: 2
configs/dcn/faster_rcnn_dcn_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: FasterRCNN
max_iters: 90000
use_gpu: true
snapshot_iter: 10000
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
metric: COCO
weights: output/faster_rcnn_dcn_r50_fpn_1x/model_final
num_classes: 81
FasterRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNet:
depth: 50
norm_type: bn
feature_maps: [2, 3, 4, 5]
freeze_at: 2
dcn_v2_stages: [3, 4, 5]
FPN:
min_level: 2
max_level: 6
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
min_level: 2
max_level: 6
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_positive_overlap: 0.7
rpn_negative_overlap: 0.3
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
min_level: 2
max_level: 5
box_resolution: 7
sampling_ratio: 2
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_lo: 0.0
bg_thresh_hi: 0.5
fg_fraction: 0.25
fg_thresh: 0.5
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [60000, 80000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../faster_fpn_reader.yml'
TrainReader:
# batch size per device
batch_size: 2
configs/dcn/faster_rcnn_dcn_r50_vd_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: FasterRCNN
max_iters: 180000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_pretrained.tar
weights: output/faster_rcnn_dcn_r50_vd_fpn_2x/model_final
metric: COCO
num_classes: 81
FasterRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNet:
depth: 50
feature_maps: [2, 3, 4, 5]
freeze_at: 2
norm_type: bn
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 2000
pre_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 1000
pre_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
box_resolution: 7
sampling_ratio: 2
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_hi: 0.5
bg_thresh_lo: 0.0
fg_fraction: 0.25
fg_thresh: 0.5
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.02
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [120000, 160000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../faster_fpn_reader.yml'
TrainReader:
# batch size per device
batch_size: 2
configs/dcn/faster_rcnn_dcn_x101_vd_64x4d_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: FasterRCNN
max_iters: 180000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNeXt101_vd_64x4d_pretrained.tar
weights: output/faster_rcnn_dcn_x101_vd_64x4d_fpn_1x/model_final
metric: COCO
num_classes: 81
FasterRCNN:
backbone: ResNeXt
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNeXt:
depth: 101
feature_maps: [2, 3, 4, 5]
freeze_at: 2
group_width: 4
groups: 64
norm_type: bn
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
anchor_sizes: [32, 64, 128, 256, 512]
aspect_ratios: [0.5, 1.0, 2.0]
stride: [16.0, 16.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 2000
pre_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
post_nms_top_n: 1000
pre_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
box_resolution: 7
sampling_ratio: 2
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_hi: 0.5
bg_thresh_lo: 0.0
fg_fraction: 0.25
fg_thresh: 0.5
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [120000, 160000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../faster_fpn_reader.yml'
configs/dcn/mask_rcnn_dcn_r101_vd_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: MaskRCNN
max_iters: 180000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet101_vd_pretrained.tar
weights: output/mask_rcnn_dcn_r101_vd_fpn_1x/model_final
metric: COCO
num_classes: 81
MaskRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNet:
depth: 101
feature_maps: [2, 3, 4, 5]
freeze_at: 2
norm_type: bn
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
aspect_ratios: [0.5, 1.0, 2.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
sampling_ratio: 2
box_resolution: 7
mask_resolution: 14
MaskHead:
dilation: 1
conv_dim: 256
num_convs: 4
resolution: 28
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_hi: 0.5
bg_thresh_lo: 0.0
fg_fraction: 0.25
fg_thresh: 0.5
MaskAssigner:
resolution: 28
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [120000, 160000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../mask_fpn_reader.yml'
configs/dcn/mask_rcnn_dcn_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: MaskRCNN
use_gpu: true
max_iters: 180000
snapshot_iter: 10000
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
metric: COCO
weights: output/mask_rcnn_dcn_r50_fpn_1x/model_final
num_classes: 81
MaskRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNet:
depth: 50
feature_maps: [2, 3, 4, 5]
freeze_at: 2
norm_type: bn
dcn_v2_stages: [3, 4, 5]
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
aspect_ratios: [0.5, 1.0, 2.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
sampling_ratio: 2
box_resolution: 7
mask_resolution: 14
MaskHead:
dilation: 1
conv_dim: 256
num_convs: 4
resolution: 28
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_hi: 0.5
bg_thresh_lo: 0.0
fg_fraction: 0.25
fg_thresh: 0.5
MaskAssigner:
resolution: 28
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [120000, 160000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../mask_fpn_reader.yml'
configs/dcn/mask_rcnn_dcn_r50_vd_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: MaskRCNN
use_gpu: true
max_iters: 360000
snapshot_iter: 10000
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_pretrained.tar
metric: COCO
weights: output/mask_rcnn_dcn_r50_vd_fpn_2x/model_final
num_classes: 81
MaskRCNN:
backbone: ResNet
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNet:
depth: 50
feature_maps: [2, 3, 4, 5]
freeze_at: 2
norm_type: bn
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
aspect_ratios: [0.5, 1.0, 2.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
box_resolution: 7
sampling_ratio: 2
mask_resolution: 14
MaskHead:
dilation: 1
conv_dim: 256
num_convs: 4
resolution: 28
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_hi: 0.5
bg_thresh_lo: 0.0
fg_fraction: 0.25
fg_thresh: 0.5
MaskAssigner:
resolution: 28
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [240000, 320000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../mask_fpn_reader.yml'
configs/dcn/mask_rcnn_dcn_x101_vd_64x4d_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
architecture: MaskRCNN
max_iters: 180000
snapshot_iter: 10000
use_gpu: true
log_smooth_window: 20
log_iter: 20
save_dir: output
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNeXt101_vd_64x4d_pretrained.tar
weights: output/mask_rcnn_dcn_x101_vd_64x4d_fpn_1x/model_final
metric: COCO
num_classes: 81
MaskRCNN:
backbone: ResNeXt
fpn: FPN
rpn_head: FPNRPNHead
roi_extractor: FPNRoIAlign
bbox_head: BBoxHead
bbox_assigner: BBoxAssigner
ResNeXt:
depth: 101
feature_maps: [2, 3, 4, 5]
freeze_at: 2
group_width: 4
groups: 64
norm_type: bn
variant: d
dcn_v2_stages: [3, 4, 5]
FPN:
max_level: 6
min_level: 2
num_chan: 256
spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
FPNRPNHead:
anchor_generator:
aspect_ratios: [0.5, 1.0, 2.0]
variance: [1.0, 1.0, 1.0, 1.0]
anchor_start_size: 32
max_level: 6
min_level: 2
num_chan: 256
rpn_target_assign:
rpn_batch_size_per_im: 256
rpn_fg_fraction: 0.5
rpn_negative_overlap: 0.3
rpn_positive_overlap: 0.7
rpn_straddle_thresh: 0.0
train_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 2000
post_nms_top_n: 2000
test_proposal:
min_size: 0.0
nms_thresh: 0.7
pre_nms_top_n: 1000
post_nms_top_n: 1000
FPNRoIAlign:
canconical_level: 4
canonical_size: 224
max_level: 5
min_level: 2
sampling_ratio: 2
box_resolution: 7
mask_resolution: 14
MaskHead:
dilation: 1
conv_dim: 256
num_convs: 4
resolution: 28
BBoxAssigner:
batch_size_per_im: 512
bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
bg_thresh_hi: 0.5
bg_thresh_lo: 0.0
fg_fraction: 0.25
fg_thresh: 0.5
MaskAssigner:
resolution: 28
BBoxHead:
head: TwoFCHead
nms:
keep_top_k: 100
nms_threshold: 0.5
score_threshold: 0.05
TwoFCHead:
mlp_dim: 1024
LearningRate:
base_lr: 0.01
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones: [120000, 160000]
- !LinearWarmup
start_factor: 0.1
steps: 1000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0001
type: L2
_READER_: '../mask_fpn_reader.yml'
configs/dcn/yolov3_enhance_reader.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
TrainReader:
inputs_def:
fields: ['image', 'gt_bbox', 'gt_class', 'gt_score']
num_max_boxes: 50
use_fine_grained_loss: true
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: True
- !RandomCrop {}
- !RandomFlipImage
is_normalized: false
- !NormalizeBox {}
- !PadBox
num_max_boxes: 50
- !BboxXYXY2XYWH {}
batch_transforms:
- !RandomShape
sizes: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608]
random_inter: True
- !NormalizeImage
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
is_scale: False
is_channel_first: false
- !Permute
to_bgr: false
channel_first: True
# Gt2YoloTarget is only used when use_fine_grained_loss set as true,
# this operator will be deleted automatically if use_fine_grained_loss
# is set as false
- !Gt2YoloTarget
anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
anchors: [[10, 13], [16, 30], [33, 23],
[30, 61], [62, 45], [59, 119],
[116, 90], [156, 198], [373, 326]]
downsample_ratios: [32, 16, 8]
batch_size: 8
shuffle: true
drop_last: true
worker_num: 8
bufsize: 16
use_process: true
EvalReader:
inputs_def:
image_shape: [3, 608, 608]
fields: ['image', 'im_size', 'im_id']
num_max_boxes: 50
dataset:
!COCODataSet
dataset_dir: dataset/coco
anno_path: annotations/instances_val2017.json
image_dir: val2017
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: True
with_mixup: false
- !ResizeImage
interp: 2
target_size: 608
- !NormalizeImage
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
is_scale: False
is_channel_first: false
- !PadBox
num_max_boxes: 50
- !Permute
to_bgr: false
channel_first: True
batch_size: 8
drop_empty: false
worker_num: 8
bufsize: 16
TestReader:
inputs_def:
image_shape: [3, 608, 608]
fields: ['image', 'im_size', 'im_id']
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
with_background: false
sample_transforms:
- !DecodeImage
to_rgb: True
with_mixup: false
- !ResizeImage
interp: 2
target_size: 608
- !NormalizeImage
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
is_scale: False
is_channel_first: false
- !Permute
to_bgr: false
channel_first: True
batch_size: 1
configs/dcn/yolov3_r50vd_dcn.yml 已删除 100755 → 0
浏览文件 @ 7cf31190
architecture: YOLOv3
use_gpu: true
max_iters: 500000
log_smooth_window: 20
save_dir: output
snapshot_iter: 20000
metric: COCO
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_pretrained.tar
weights: output/yolov3_r50vd_dcn/model_final
num_classes: 80
use_fine_grained_loss: false
YOLOv3:
backbone: ResNet
yolo_head: YOLOv3Head
ResNet:
norm_type: sync_bn
freeze_at: 0
freeze_norm: false
norm_decay: 0.
depth: 50
feature_maps: [3, 4, 5]
variant: d
dcn_v2_stages: [5]
YOLOv3Head:
anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
anchors: [[10, 13], [16, 30], [33, 23],
[30, 61], [62, 45], [59, 119],
[116, 90], [156, 198], [373, 326]]
norm_decay: 0.
yolo_loss: YOLOv3Loss
nms:
background_label: -1
keep_top_k: 100
nms_threshold: 0.45
nms_top_k: 1000
normalized: false
score_threshold: 0.01
YOLOv3Loss:
# batch_size here is only used for fine grained loss, not used
# for training batch_size setting, training batch_size setting
# is in configs/yolov3_reader.yml TrainReader.batch_size, batch
# size here should be set as same value as TrainReader.batch_size
batch_size: 8
ignore_thresh: 0.7
label_smooth: false
LearningRate:
base_lr: 0.001
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones:
- 400000
- 450000
- !LinearWarmup
start_factor: 0.
steps: 4000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0005
type: L2
_READER_: '../yolov3_reader.yml'
configs/dcn/yolov3_r50vd_dcn_db_iouaware_obj365_pretrained_coco.yml 已删除 100755 → 0
浏览文件 @ 7cf31190
architecture: YOLOv3
use_gpu: true
max_iters: 85000
log_smooth_window: 1
save_dir: output
snapshot_iter: 10000
metric: COCO
pretrain_weights: https://paddlemodels.bj.bcebos.com/object_detection/ResNet50_vd_dcn_db_obj365_pretrained.tar
weights: output/yolov3_r50vd_dcn_db_iouaware_obj365_pretrained_coco/model_final
num_classes: 80
use_fine_grained_loss: true
YOLOv3:
backbone: ResNet
yolo_head: YOLOv3Head
use_fine_grained_loss: true
ResNet:
norm_type: sync_bn
freeze_at: 0
freeze_norm: false
norm_decay: 0.
depth: 50
feature_maps: [3, 4, 5]
variant: d
dcn_v2_stages: [5]
YOLOv3Head:
anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
anchors: [[10, 13], [16, 30], [33, 23],
[30, 61], [62, 45], [59, 119],
[116, 90], [156, 198], [373, 326]]
norm_decay: 0.
iou_aware: true
iou_aware_factor: 0.4
yolo_loss: YOLOv3Loss
nms:
background_label: -1
keep_top_k: 100
nms_threshold: 0.45
nms_top_k: 1000
normalized: false
score_threshold: 0.01
drop_block: true
YOLOv3Loss:
batch_size: 8
ignore_thresh: 0.7
label_smooth: false
use_fine_grained_loss: true
iou_loss: IouLoss
iou_aware_loss: IouAwareLoss
IouLoss:
loss_weight: 2.5
max_height: 608
max_width: 608
IouAwareLoss:
loss_weight: 1.0
max_height: 608
max_width: 608
LearningRate:
base_lr: 0.001
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones:
- 55000
- 75000
- !LinearWarmup
start_factor: 0.
steps: 4000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0005
type: L2
_READER_: 'yolov3_enhance_reader.yml'
configs/dcn/yolov3_r50vd_dcn_db_iouloss_obj365_pretrained_coco.yml 已删除 100755 → 0
浏览文件 @ 7cf31190
architecture: YOLOv3
use_gpu: true
max_iters: 85000
log_smooth_window: 20
save_dir: output
snapshot_iter: 10000
metric: COCO
pretrain_weights: https://paddlemodels.bj.bcebos.com/object_detection/ResNet50_vd_dcn_db_obj365_pretrained.tar
weights: output/yolov3_r50vd_dcn_db_iouloss_obj365_pretrained_coco/model_final
num_classes: 80
use_fine_grained_loss: true
YOLOv3:
backbone: ResNet
yolo_head: YOLOv3Head
use_fine_grained_loss: true
ResNet:
norm_type: sync_bn
freeze_at: 0
freeze_norm: false
norm_decay: 0.
depth: 50
feature_maps: [3, 4, 5]
variant: d
dcn_v2_stages: [5]
YOLOv3Head:
anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
anchors: [[10, 13], [16, 30], [33, 23],
[30, 61], [62, 45], [59, 119],
[116, 90], [156, 198], [373, 326]]
norm_decay: 0.
yolo_loss: YOLOv3Loss
nms:
background_label: -1
keep_top_k: 100
nms_threshold: 0.45
nms_top_k: 1000
normalized: false
score_threshold: 0.01
drop_block: true
YOLOv3Loss:
# batch_size here is only used for fine grained loss, not used
# for training batch_size setting, training batch_size setting
# is in configs/yolov3_reader.yml TrainReader.batch_size, batch
# size here should be set as same value as TrainReader.batch_size
batch_size: 8
ignore_thresh: 0.7
label_smooth: false
use_fine_grained_loss: true
iou_loss: IouLoss
IouLoss:
loss_weight: 2.5
max_height: 608
max_width: 608
LearningRate:
base_lr: 0.001
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones:
- 55000
- 75000
- !LinearWarmup
start_factor: 0.
steps: 4000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0005
type: L2
_READER_: 'yolov3_enhance_reader.yml'
configs/dcn/yolov3_r50vd_dcn_db_obj365_pretrained_coco.yml 已删除 100755 → 0
浏览文件 @ 7cf31190
architecture: YOLOv3
use_gpu: true
max_iters: 85000
log_smooth_window: 20
save_dir: output
snapshot_iter: 10000
metric: COCO
pretrain_weights: https://paddlemodels.bj.bcebos.com/object_detection/ResNet50_vd_dcn_db_obj365_pretrained.tar
weights: output/yolov3_r50vd_dcn_db_obj365_pretrained_coco/model_final
num_classes: 80
use_fine_grained_loss: true
YOLOv3:
backbone: ResNet
yolo_head: YOLOv3Head
use_fine_grained_loss: true
ResNet:
norm_type: sync_bn
freeze_at: 0
freeze_norm: false
norm_decay: 0.
depth: 50
feature_maps: [3, 4, 5]
variant: d
dcn_v2_stages: [5]
YOLOv3Head:
anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
anchors: [[10, 13], [16, 30], [33, 23],
[30, 61], [62, 45], [59, 119],
[116, 90], [156, 198], [373, 326]]
norm_decay: 0.
yolo_loss: YOLOv3Loss
nms:
background_label: -1
keep_top_k: 100
nms_threshold: 0.45
nms_top_k: 1000
normalized: false
score_threshold: 0.01
drop_block: true
keep_prob: 0.94
YOLOv3Loss:
# batch_size here is only used for fine grained loss, not used
# for training batch_size setting, training batch_size setting
# is in configs/yolov3_reader.yml TrainReader.batch_size, batch
# size here should be set as same value as TrainReader.batch_size
batch_size: 8
ignore_thresh: 0.7
label_smooth: false
use_fine_grained_loss: true
LearningRate:
base_lr: 0.001
schedulers:
- !PiecewiseDecay
gamma: 0.1
milestones:
- 55000
- 75000
- !LinearWarmup
start_factor: 0.
steps: 4000
OptimizerBuilder:
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.0005
type: L2
_READER_: 'yolov3_enhance_reader.yml'
configs/efficientdet_d0.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/face_detection/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
**文档教程请参考：** [FACE_DETECTION.md](../../docs/featured_model/FACE_DETECTION.md) <br/>
**English document please refer:** [FACE_DETECTION_en.md](../../docs/featured_model/FACE_DETECTION_en.md)
configs/face_detection/blazeface.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/face_detection/blazeface_nas.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/face_detection/blazeface_nas_v2.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/face_detection/faceboxes.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/face_detection/faceboxes_lite.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_fpn_reader.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_cbr101_vd_dual_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_cbr50_vd_dual_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r101_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r101_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r101_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r101_vd_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r101_vd_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r50_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r50_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r50_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r50_vd_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_r50_vd_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_se154_vd_fpn_s1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_x101_vd_64x4d_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_rcnn_x101_vd_64x4d_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/faster_reader.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/gcnet/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
# GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond
## Introduction
- GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond
: [https://arxiv.org/abs/1904.11492](https://arxiv.org/abs/1904.11492)
```
@article{DBLP:journals/corr/abs-1904-11492,
author = {Yue Cao and
Jiarui Xu and
Stephen Lin and
Fangyun Wei and
Han Hu},
title = {GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond},
journal = {CoRR},
volume = {abs/1904.11492},
year = {2019},
url = {http://arxiv.org/abs/1904.11492},
archivePrefix = {arXiv},
eprint = {1904.11492},
timestamp = {Tue, 09 Jul 2019 16:48:55 +0200},
biburl = {https://dblp.org/rec/bib/journals/corr/abs-1904-11492},
bibsource = {dblp computer science bibliography, https://dblp.org}
}
```
## Model Zoo
| Backbone | Type | Context| Image/gpu | Lr schd | Inf time (fps) | Box AP | Mask AP | Download | Configs |
| :---------------------- | :-------------: | :-------------: | :-------: | :-----: | :------------: | :----: | :-----: | :----------------------------------------------------------: | :-----: |
| ResNet50-vd-FPN | Mask | GC(c3-c5, r16, add) | 2 | 2x | 15.31 | 41.4 | 36.8 | [model](https://paddlemodels.bj.bcebos.com/object_detection/mask_rcnn_r50_vd_fpn_gcb_add_r16_2x.tar) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/gcnet/mask_rcnn_r50_vd_fpn_gcb_add_r16_2x.yml) |
| ResNet50-vd-FPN | Mask | GC(c3-c5, r16, mul) | 2 | 2x | 15.35 | 40.7 | 36.1 | [model](https://paddlemodels.bj.bcebos.com/object_detection/mask_rcnn_r50_vd_fpn_gcb_mul_r16_2x.tar) | [config](https://github.com/PaddlePaddle/PaddleDetection/tree/master/configs/gcnet/mask_rcnn_r50_vd_fpn_gcb_mul_r16_2x.yml) |
configs/gcnet/README_cn.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/gn/faster_rcnn_r50_fpn_gn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/gn/mask_rcnn_r50_fpn_gn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/hrnet/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/hrnet/faster_rcnn_hrnetv2p_w18_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/hrnet/faster_rcnn_hrnetv2p_w18_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/iou_loss/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/iou_loss/README_cn.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/libra_rcnn/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/libra_rcnn/README_cn.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_fpn_reader.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_r101_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_r101_vd_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_r50_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_r50_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_r50_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_r50_vd_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_se154_vd_fpn_s1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_x101_vd_64x4d_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_rcnn_x101_vd_64x4d_fpn_2x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mask_reader.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mobile/README.md 已删除 100755 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mobile/README_en.md 已删除 100755 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mobile/ssdlite_mobilenet_v3_large.yml 已删除 120000 → 0
浏览文件 @ 7cf31190
../ssd/ssdlite_mobilenet_v3_large.yml
\ No newline at end of file
configs/mobile/ssdlite_mobilenet_v3_small.yml 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/mobile/yolov3_mobilenet_v3.yml 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/rcnn_enhance/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/rcnn_enhance/README_en.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/res2net/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/retinanet_r101_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/retinanet_r50_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/retinanet_x101_vd_64x4d_fpn_1x.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/ssd/ssd_mobilenet_v1_voc.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/ssd/ssd_vgg16_300.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/ssd/ssd_vgg16_300_voc.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/ssd/ssd_vgg16_512.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/ssd/ssd_vgg16_512_voc.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/ssd/ssdlite_mobilenet_v1.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/ssd/ssdlite_mobilenet_v3_large.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/ssd/ssdlite_mobilenet_v3_small.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_darknet.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_darknet_voc.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_darknet_voc_diouloss.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_mobilenet_v1.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_mobilenet_v1_fruit.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_mobilenet_v1_voc.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_mobilenet_v3.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_r34.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_r34_voc.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov3_reader.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov4/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov4/yolov4_cspdarknet.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov4/yolov4_cspdarknet_coco.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
configs/yolov4/yolov4_cspdarknet_voc.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/PedestrianDetection/demo/001.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/PedestrianDetection/demo/002.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/PedestrianDetection/demo/003.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/PedestrianDetection/demo/004.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/PedestrianDetection/pedestrian.json 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/VehicleDetection/demo/001.jpeg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/VehicleDetection/demo/003.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/VehicleDetection/demo/004.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/VehicleDetection/demo/005.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
contrib/VehicleDetection/vehicle.json 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
demo/000000014439.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
demo/000000014439_640x640.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
demo/000000087038.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
demo/000000570688.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
demo/infer_cfg.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
demo/mask_rcnn_demo.ipynb 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
demo/orange_71.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
deploy/cpp/CMakeSettings.json 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/CHANGELOG.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/FAQ.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/MODEL_ZOO.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/MODEL_ZOO_cn.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/Makefile 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/MODEL_TECHNICAL.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/READER.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/TRANSFER_LEARNING.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/config_doc/CONFIG.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/config_doc/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/deploy/DEPLOY_CPP.md 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/deploy/DEPLOY_PY.md 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/deploy/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/slim/MODEL_ZOO.md 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/slim/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/slim/nas/NAS.md 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/slim/nas/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/slim/prune/PRUNE.md 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/advanced_tutorials/slim/prune/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/conf.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/ANCHOR_FREE_DETECTION.md 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/CONTRIB.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/CONTRIB_cn.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/FACE_DETECTION.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/FACE_DETECTION_en.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/MOBILE_SIDE.md 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/SERVER_SIDE.md 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/YOLO_V4.md 已删除 120000 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/YOLOv3_ENHANCEMENT.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/champion_model/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/featured_model/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/000000087038_res.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/000000570688.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/PedestrianDetection_001.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/PedestrianDetection_004.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/VehicleDetection_001.jpeg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/VehicleDetection_005.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/bench_ssd_yolo_infer.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/cas.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/dropblock.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/map_fps.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/gcnet_gcblock_module.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/gcnet_snl_module.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/gcnet_snl_module_simple.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/gcnet_snl_out.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/gcnet_snl_out_simple.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/iou_loss_diou_bbox_loss.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/iou_loss_diou_ciou_nms.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/iou_loss_diou_diou_final.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/iou_loss_giou_calc.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/iou_loss_giou_pipeline.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/libra_rcnn_libraloss_equ.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/libra_rcnn_loss_grad.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/libra_rcnn_pipeline.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models/libra_rcnn_smooth_l1_equ.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/models_figure.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/obj365_gt.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/obj365_pred.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/oidv5_gt.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/oidv5_model_framework.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/oidv5_pred.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/orange_71_detection.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/pssdet.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/reader_figure.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/images/visualdl_fruit.jpg 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/make.bat 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/requirements.txt 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/tutorials/GETTING_STARTED.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/tutorials/GETTING_STARTED_cn.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/tutorials/INSTALL.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/tutorials/INSTALL_cn.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/tutorials/QUICK_STARTED.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/tutorials/QUICK_STARTED_cn.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
docs/tutorials/index.rst 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/experimental/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/experimental/mixed_precision.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/anchor_heads/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/anchor_heads/corner_head.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/anchor_heads/efficient_head.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/anchor_heads/fcos_head.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/anchor_heads/iou_aware.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/anchor_heads/retina_head.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/anchor_heads/rpn_head.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/anchor_heads/yolo_head.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/blazeface.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/cascade_rcnn.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/efficientdet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/faceboxes.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/faster_rcnn.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/fcos.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/input_helper.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/mask_rcnn.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/retinanet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/ssd.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/architectures/yolo.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/bfp.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/bifpn.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/blazenet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/cb_resnet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/cspdarknet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/darknet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/efficientnet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/faceboxnet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/fpn.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/gc_block.py 已删除 100755 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/hourglass.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/hrfpn.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/hrnet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/mobilenet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/mobilenet_v3.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/name_adapter.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/nonlocal_helper.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/res2net.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/resnet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/resnext.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/senet.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/backbones/vgg.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/balanced_l1_loss.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/diou_loss.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/diou_loss_yolo.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/fcos_loss.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/giou_loss.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/iou_aware_loss.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/iou_loss.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/smooth_l1_loss.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/losses/yolo_loss.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/ops.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/roi_extractors/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/roi_heads/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/roi_heads/bbox_head.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/roi_heads/cascade_head.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/roi_heads/mask_head.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/target_assigners.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/tests/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/tests/decorator_helper.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
ppdet/modeling/tests/test_architectures.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/distillation/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/distillation/distill.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/nas/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/nas/blazeface.yml 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/nas/latency_855.txt 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/nas/search_space/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/nas/search_space/blazefacespace_nas.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/nas/train_nas.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/prune/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/prune/eval.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/prune/export_model.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/prune/prune.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/quantization/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/quantization/eval.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/quantization/export_model.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/quantization/images/ConvertToInt8Pass.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/quantization/images/FreezePass.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/quantization/images/TransformPass.png 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/quantization/infer.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/quantization/train.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/sensitive/README.md 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
slim/sensitive/sensitive.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
tools/__init__.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
tools/configure.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
tools/cpp_infer.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
tools/eval.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
tools/export_model.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
tools/export_serving_model.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
tools/face_eval.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
tools/infer.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
tools/train.py 已删除 100644 → 0
浏览文件 @ 7cf31190
此差异已折叠。
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或