提交 9790444e 编写于 作者: littletomatodonkey's avatar littletomatodonkey

add resnest fast config and fix flops

上级 7505a78b
mode: 'train'
ARCHITECTURE:
name: 'ResNeSt50_fast_1s1x64d'
pretrained_model: ""
model_save_dir: "./output/"
classes_num: 1000
total_images: 1281167
save_interval: 1
validate: True
valid_interval: 1
epochs: 300
topk: 5
image_shape: [3, 224, 224]
use_mix: True
ls_epsilon: 0.1
LEARNING_RATE:
function: 'CosineWarmup'
params:
lr: 0.1
OPTIMIZER:
function: 'Momentum'
params:
momentum: 0.9
regularizer:
function: 'L2'
factor: 0.000070
TRAIN:
batch_size: 256
num_workers: 4
file_list: "./dataset/ILSVRC2012/train_list.txt"
data_dir: "./dataset/ILSVRC2012/"
shuffle_seed: 0
transforms:
- DecodeImage:
to_rgb: True
to_np: False
channel_first: False
- RandCropImage:
size: 224
- RandFlipImage:
flip_code: 1
- AutoAugment:
- NormalizeImage:
scale: 1./255.
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
order: ''
- ToCHWImage:
mix:
- CutmixOperator:
alpha: 0.2
VALID:
batch_size: 64
num_workers: 4
file_list: "./dataset/ILSVRC2012/val_list.txt"
data_dir: "./dataset/ILSVRC2012/"
shuffle_seed: 0
transforms:
- DecodeImage:
to_rgb: True
to_np: False
channel_first: False
- ResizeImage:
resize_short: 256
- CropImage:
size: 224
- NormalizeImage:
scale: 1.0/255.0
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
order: ''
- ToCHWImage:
...@@ -58,9 +58,9 @@ Currently there are 32 pretrained models of the mobile series open source by Pad ...@@ -58,9 +58,9 @@ Currently there are 32 pretrained models of the mobile series open source by Pad
| ShuffleNetV2_x1_5 | 0.716 | 0.902 | 0.726 | | 0.580 | 3.470 | | ShuffleNetV2_x1_5 | 0.716 | 0.902 | 0.726 | | 0.580 | 3.470 |
| ShuffleNetV2_x2_0 | 0.732 | 0.912 | 0.749 | | 1.120 | 7.320 | | ShuffleNetV2_x2_0 | 0.732 | 0.912 | 0.749 | | 1.120 | 7.320 |
| ShuffleNetV2_swish | 0.700 | 0.892 | | | 0.290 | 2.260 | | ShuffleNetV2_swish | 0.700 | 0.892 | | | 0.290 | 2.260 |
| GhostNet_x0_5 | 0.668 | 0.869 | 0.662 | 0.866 | 0.041 | 2.600 | | GhostNet_x0_5 | 0.668 | 0.869 | 0.662 | 0.866 | 0.082 | 2.600 |
| GhostNet_x1_0 | 0.740 | 0.916 | 0.739 | 0.914 | 0.147 | 5.200 | | GhostNet_x1_0 | 0.740 | 0.916 | 0.739 | 0.914 | 0.294 | 5.200 |
| GhostNet_x1_3 | 0.757 | 0.925 | 0.757 | 0.927 | 0.220 | 7.300 | | GhostNet_x1_3 | 0.757 | 0.925 | 0.757 | 0.927 | 0.440 | 7.300 |
## Inference speed and storage size based on SD855 ## Inference speed and storage size based on SD855
......
...@@ -6,4 +6,5 @@ The ResNeSt series was proposed in 2020. The original resnet network structure h ...@@ -6,4 +6,5 @@ The ResNeSt series was proposed in 2020. The original resnet network structure h
| Models | Top1 | Top5 | Reference<br>top1 | Reference<br>top5 | FLOPS<br>(G) | Parameters<br>(M) | | Models | Top1 | Top5 | Reference<br>top1 | Reference<br>top5 | FLOPS<br>(G) | Parameters<br>(M) |
|:--:|:--:|:--:|:--:|:--:|:--:|:--:| |:--:|:--:|:--:|:--:|:--:|:--:|:--:|
| ResNeSt50 | 0.8102 | 0.9542| 0.8113 | -|5.39 | 27.5 | | ResNeSt50_fast_1s1x64d | 0.8035 | 0.9528| 0.8035 | -| 8.68 | 26.3 |
| ResNeSt50 | 0.8102 | 0.9542| 0.8113 | -| 10.78 | 27.5 |
...@@ -191,6 +191,13 @@ python tools/infer/predict.py \ ...@@ -191,6 +191,13 @@ python tools/infer/predict.py \
- [Fix_ResNeXt101_32x48d_wsl](https://paddle-imagenet-models-name.bj.bcebos.com/Fix_ResNeXt101_32x48d_wsl_pretrained.tar) - [Fix_ResNeXt101_32x48d_wsl](https://paddle-imagenet-models-name.bj.bcebos.com/Fix_ResNeXt101_32x48d_wsl_pretrained.tar)
- ResNeSt and RegNet series
- ResNeSt系列<sup>[[24](#ref24)]</sup>([paper link](https://arxiv.org/abs/2004.08955))
- [ResNeSt50_fast_1s1x64d)(https://paddle-imagenet-models-name.bj.bcebos.com/ResNeSt50_fast_1s1x64d_pretrained.tar)
- [ResNeSt50)(https://paddle-imagenet-models-name.bj.bcebos.com/ResNeSt50_pretrained.tar)
- Other models - Other models
- AlexNet series<sup>[[18](#ref18)]</sup>([paper link](https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf)) - AlexNet series<sup>[[18](#ref18)]</sup>([paper link](https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf))
- [AlexNet](https://paddle-imagenet-models-name.bj.bcebos.com/AlexNet_pretrained.tar) - [AlexNet](https://paddle-imagenet-models-name.bj.bcebos.com/AlexNet_pretrained.tar)
...@@ -261,3 +268,5 @@ python tools/infer/predict.py \ ...@@ -261,3 +268,5 @@ python tools/infer/predict.py \
<a name="ref22">[22]</a> Ding X, Guo Y, Ding G, et al. Acnet: Strengthening the kernel skeletons for powerful cnn via asymmetric convolution blocks[C]//Proceedings of the IEEE International Conference on Computer Vision. 2019: 1911-1920. <a name="ref22">[22]</a> Ding X, Guo Y, Ding G, et al. Acnet: Strengthening the kernel skeletons for powerful cnn via asymmetric convolution blocks[C]//Proceedings of the IEEE International Conference on Computer Vision. 2019: 1911-1920.
<a name="ref23">[23]</a> Han K, Wang Y, Tian Q, et al. GhostNet: More features from cheap operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020: 1580-1589. <a name="ref23">[23]</a> Han K, Wang Y, Tian Q, et al. GhostNet: More features from cheap operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020: 1580-1589.
<a name="ref24">[24]</a> Zhang H, Wu C, Zhang Z, et al. Resnest: Split-attention networks[J]. arXiv preprint arXiv:2004.08955, 2020.
...@@ -59,9 +59,9 @@ GhosttNet是华为于2020年提出的一种全新的轻量化网络结构,通 ...@@ -59,9 +59,9 @@ GhosttNet是华为于2020年提出的一种全新的轻量化网络结构,通
| ShuffleNetV2_x1_5 | 0.716 | 0.902 | 0.726 | | 0.580 | 3.470 | | ShuffleNetV2_x1_5 | 0.716 | 0.902 | 0.726 | | 0.580 | 3.470 |
| ShuffleNetV2_x2_0 | 0.732 | 0.912 | 0.749 | | 1.120 | 7.320 | | ShuffleNetV2_x2_0 | 0.732 | 0.912 | 0.749 | | 1.120 | 7.320 |
| ShuffleNetV2_swish | 0.700 | 0.892 | | | 0.290 | 2.260 | | ShuffleNetV2_swish | 0.700 | 0.892 | | | 0.290 | 2.260 |
| GhostNet_x0_5 | 0.668 | 0.869 | 0.662 | 0.866 | 0.041 | 2.600 | | GhostNet_x0_5 | 0.668 | 0.869 | 0.662 | 0.866 | 0.082 | 2.600 |
| GhostNet_x1_0 | 0.740 | 0.916 | 0.739 | 0.914 | 0.147 | 5.200 | | GhostNet_x1_0 | 0.740 | 0.916 | 0.739 | 0.914 | 0.294 | 5.200 |
| GhostNet_x1_3 | 0.757 | 0.925 | 0.757 | 0.927 | 0.220 | 7.300 | | GhostNet_x1_3 | 0.757 | 0.925 | 0.757 | 0.927 | 0.440 | 7.300 |
## 基于SD855的预测速度和存储大小 ## 基于SD855的预测速度和存储大小
......
...@@ -9,6 +9,5 @@ ResNeSt系列模型是在2020年提出的,在原有的resnet网络结构上做 ...@@ -9,6 +9,5 @@ ResNeSt系列模型是在2020年提出的,在原有的resnet网络结构上做
| Models | Top1 | Top5 | Reference<br>top1 | Reference<br>top5 | FLOPS<br>(G) | Parameters<br>(M) | | Models | Top1 | Top5 | Reference<br>top1 | Reference<br>top5 | FLOPS<br>(G) | Parameters<br>(M) |
|:--:|:--:|:--:|:--:|:--:|:--:|:--:| |:--:|:--:|:--:|:--:|:--:|:--:|:--:|
| ResNeSt50 | 0.8102 | 0.9542| 0.8113 | -|5.39 | 27.5 | | ResNeSt50_fast_1s1x64d | 0.8035 | 0.9528| 0.8035 | -| 8.68 | 26.3 |
| ResNeSt50 | 0.8102 | 0.9542| 0.8113 | -| 10.78 | 27.5 |
...@@ -190,6 +190,11 @@ python tools/infer/predict.py \ ...@@ -190,6 +190,11 @@ python tools/infer/predict.py \
- [ResNeXt101_32x48d_wsl](https://paddle-imagenet-models-name.bj.bcebos.com/ResNeXt101_32x48d_wsl_pretrained.tar) - [ResNeXt101_32x48d_wsl](https://paddle-imagenet-models-name.bj.bcebos.com/ResNeXt101_32x48d_wsl_pretrained.tar)
- [Fix_ResNeXt101_32x48d_wsl](https://paddle-imagenet-models-name.bj.bcebos.com/Fix_ResNeXt101_32x48d_wsl_pretrained.tar) - [Fix_ResNeXt101_32x48d_wsl](https://paddle-imagenet-models-name.bj.bcebos.com/Fix_ResNeXt101_32x48d_wsl_pretrained.tar)
- ResNeSt与RegNet系列
- ResNeSt系列<sup>[[24](#ref24)]</sup>([论文地址](https://arxiv.org/abs/2004.08955))
- [ResNeSt50_fast_1s1x64d)(https://paddle-imagenet-models-name.bj.bcebos.com/ResNeSt50_fast_1s1x64d_pretrained.tar)
- [ResNeSt50)(https://paddle-imagenet-models-name.bj.bcebos.com/ResNeSt50_pretrained.tar)
- 其他模型 - 其他模型
- AlexNet系列<sup>[[18](#ref18)]</sup>([论文地址](https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf)) - AlexNet系列<sup>[[18](#ref18)]</sup>([论文地址](https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf))
...@@ -261,3 +266,5 @@ python tools/infer/predict.py \ ...@@ -261,3 +266,5 @@ python tools/infer/predict.py \
<a name="ref22">[22]</a> Ding X, Guo Y, Ding G, et al. Acnet: Strengthening the kernel skeletons for powerful cnn via asymmetric convolution blocks[C]//Proceedings of the IEEE International Conference on Computer Vision. 2019: 1911-1920. <a name="ref22">[22]</a> Ding X, Guo Y, Ding G, et al. Acnet: Strengthening the kernel skeletons for powerful cnn via asymmetric convolution blocks[C]//Proceedings of the IEEE International Conference on Computer Vision. 2019: 1911-1920.
<a name="ref23">[23]</a> Han K, Wang Y, Tian Q, et al. GhostNet: More features from cheap operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020: 1580-1589. <a name="ref23">[23]</a> Han K, Wang Y, Tian Q, et al. GhostNet: More features from cheap operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020: 1580-1589.
<a name="ref24">[24]</a> Zhang H, Wu C, Zhang Z, et al. Resnest: Split-attention networks[J]. arXiv preprint arXiv:2004.08955, 2020.
Markdown is supported
0% .
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