Skip to content

P
PaddleSlim

PaddlePaddle / PaddleSlim
1 年多 前同步成功

通知 51
Star 1434
Fork 344
P
PaddleSlim
未验证 提交 1e4656fe 编写于 作者: W whs 提交者： GitHub
浏览文件
操作

add PaddleLite time cost and TensorRT FPS (#147) (#149)

上级 5668782b
显示空白变更内容
内联 并排
Showing with 67 addition and 35 deletion
docs/en/model_zoo_en.md
浏览文件 @ 1e4656fe
...@@ -144,18 +144,26 @@ Dataset：WIDER-FACE ...@@ -144,18 +144,26 @@ Dataset：WIDER-FACE
Dataset：Pasacl VOC & COCO 2017 Dataset：Pasacl VOC & COCO 2017
| Model | Method | Dataset | Image/GPU | Input 608 Box AP | Input 416 Box AP | Input 320 Box AP | Model Size(MB) | GFLOPs (608*608) | Download | PaddleLite:
| :----------------------------: | :---------------: | :--------: | :-------: | :--------------: | :--------------: | :--------------: | :------------: | :--------------: | :----------------------------------------------------------: |
| MobileNet-V1-YOLOv3 | Baseline | Pascal VOC | 8 | 76.2 | 76.7 | 75.3 | 94 | 40.49 | [model](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1_voc.tar) | env: Qualcomm SnapDragon 845 + armv8
| MobileNet-V1-YOLOv3 | sensitive -52.88% | Pascal VOC | 8 | 77.6 (+1.4) | 77.7 (1.0) | 75.5 (+0.2) | 31 | 19.08 | [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_mobilenet_v1_voc_prune.tar) |
| MobileNet-V1-YOLOv3 | - | COCO | 8 | 29.3 | 29.3 | 27.0 | 95 | 41.35 | [model](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1.tar) | criterion: time cost in Thread1/Thread2/Thread4
| MobileNet-V1-YOLOv3 | sensitive -51.77% | COCO | 8 | 26.0 (-3.3) | 25.1 (-4.2) | 22.6 (-4.4) | 32 | 19.94 | [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_mobilenet_v1_prune.tar) |
| R50-dcn-YOLOv3 | - | COCO | 8 | 39.1 | - | - | 177 | 89.60 | [model](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_r50vd_dcn.tar) | PaddleLite version: v2.3
| R50-dcn-YOLOv3 | sensitive -9.37% | COCO | 8 | 39.3 (+0.2) | - | - | 150 | 81.20 | [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_prune.tar) |
| R50-dcn-YOLOv3 | sensitive -24.68% | COCO | 8 | 37.3 (-1.8) | - | - | 113 | 67.48 | [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_prune578.tar) | | Model | Method | Dataset | Image/GPU | Input 608 Box AP | Input 416 Box AP | Input 320 Box AP | Model Size(MB) | GFLOPs (608*608) | PaddleLite cost(ms)(608*608) | TensorRT speed(FPS)(608*608) | Download |
| R50-dcn-YOLOv3 obj365_pretrain | - | COCO | 8 | 41.4 | - | - | 177 | 89.60 | [model](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_r50vd_dcn_obj365_pretrained_coco.tar) | | :----------------------------: | :---------------: | :--------: | :-------: | :--------------: | :--------------: | :--------------: | :------------: | :--------------: | :--------------: | :--------------: | :----------------------------: |
| R50-dcn-YOLOv3 obj365_pretrain | sensitive -9.37% | COCO | 8 | 40.5 (-0.9) | - | - | 150 | 81.20 | [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_obj365_pretrained_coco_prune.tar) | | MobileNet-V1-YOLOv3 | Baseline | Pascal VOC | 8 | 76.2 | 76.7 | 75.3 | 94 | 40.49 | 1238\796.943\520.101 |60.40| [model](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1_voc.tar) |
| R50-dcn-YOLOv3 obj365_pretrain | sensitive -24.68% | COCO | 8 | 37.8 (-3.3) | - | - | 113 | 67.48 | [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_obj365_pretrained_coco_prune578.tar) | | MobileNet-V1-YOLOv3 | sensitive -52.88% | Pascal VOC | 8 | 77.6 (+1.4) | 77.7 (1.0) | 75.5 (+0.2) | 31 | 19.08 | 602.497\353.759\222.427 |99.36| [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_mobilenet_v1_voc_prune.tar) |
| MobileNet-V1-YOLOv3 | - | COCO | 8 | 29.3 | 29.3 | 27.0 | 95 | 41.35 |-|-| [model](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1.tar) |
| MobileNet-V1-YOLOv3 | sensitive -51.77% | COCO | 8 | 26.0 (-3.3) | 25.1 (-4.2) | 22.6 (-4.4) | 32 | 19.94 |-|73.93| [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_mobilenet_v1_prune.tar) |
| R50-dcn-YOLOv3 | - | COCO | 8 | 39.1 | - | - | 177 | 89.60 |-|27.68| [model](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_r50vd_dcn.tar) |
| R50-dcn-YOLOv3 | sensitive -9.37% | COCO | 8 | 39.3 (+0.2) | - | - | 150 | 81.20 |-|30.08| [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_prune.tar) |
| R50-dcn-YOLOv3 | sensitive -24.68% | COCO | 8 | 37.3 (-1.8) | - | - | 113 | 67.48 |-|34.32| [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_prune578.tar) |
| R50-dcn-YOLOv3 obj365_pretrain | - | COCO | 8 | 41.4 | - | - | 177 | 89.60 |-|-| [model](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_r50vd_dcn_obj365_pretrained_coco.tar) |
| R50-dcn-YOLOv3 obj365_pretrain | sensitive -9.37% | COCO | 8 | 40.5 (-0.9) | - | - | 150 | 81.20 |-|-| [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_obj365_pretrained_coco_prune.tar) |
| R50-dcn-YOLOv3 obj365_pretrain | sensitive -24.68% | COCO | 8 | 37.8 (-3.3) | - | - | 113 | 67.48 |-|-| [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_obj365_pretrained_coco_prune578.tar) |
### 2.3 Distillation ### 2.3 Distillation
...@@ -227,8 +235,16 @@ Image segmentation model PaddleLite latency (ms), input size 769x769 ...@@ -227,8 +235,16 @@ Image segmentation model PaddleLite latency (ms), input size 769x769
### 3.2 Pruning ### 3.2 Pruning
| Model | Method | mIoU | Model Size（MB） | GFLOPs | Download | PaddleLite:
| :-------: | :---------------: | :-----------: | :--------------: | :----: | :----------------------------------------------------------: |
| fast-scnn | baseline | 69.64 | 11 | 14.41 | [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape.tar) | env: Qualcomm SnapDragon 845 + armv8
| fast-scnn | uniform -17.07% | 69.58 (-0.06) | 8.5 | 11.95 | [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape_uniform-17.tar) |
| fast-scnn | sensitive -47.60% | 66.68 (-2.96) | 5.7 | 7.55 | [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape_sensitive-47.tar) | criterion: time cost in Thread1/Thread2/Thread4
PaddleLite version: v2.3
| Model | Method | mIoU | Model Size（MB） | GFLOPs | PaddleLite cost(ms) | TensorRT speed(FPS) | Download |
| :-------: | :---------------: | :-----------: | :--------------: | :----: | :--------------: | :----: | :-------------------: |
| fast-scnn | baseline | 69.64 | 11 | 14.41 | 1226.36\682.96\415.664 |39.53| [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape.tar) |
| fast-scnn | uniform -17.07% | 69.58 (-0.06) | 8.5 | 11.95 | 1140.37\656.612\415.888 |42.01| [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape_uniform-17.tar) |
| fast-scnn | sensitive -47.60% | 66.68 (-2.96) | 5.7 | 7.55 | 866.693\494.467\291.748 |51.48| [model](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape_sensitive-47.tar) |
docs/zh_cn/model_zoo.md
浏览文件 @ 1e4656fe
...@@ -110,7 +110,6 @@ PaddleLite版本： v2.3 ...@@ -110,7 +110,6 @@ PaddleLite版本： v2.3
| Darts | - | 97.135% | 3.767 | - | | Darts | - | 97.135% | 3.767 | - |
| Darts_SA(基于Darts搜索空间) | SANAS | 97.276%(+0.141%) | 3.344(-11.2%) | - | | Darts_SA(基于Darts搜索空间) | SANAS | 97.276%(+0.141%) | 3.344(-11.2%) | - |
Note: MobileNetV2_NAS 的token是：[4, 4, 5, 1, 1, 2, 1, 1, 0, 2, 6, 2, 0, 3, 4, 5, 0, 4, 5, 5, 1, 4, 8, 0, 0]. Darts_SA的token是：[5, 5, 0, 5, 5, 10, 7, 7, 5, 7, 7, 11, 10, 12, 10, 0, 5, 3, 10, 8]. Note: MobileNetV2_NAS 的token是：[4, 4, 5, 1, 1, 2, 1, 1, 0, 2, 6, 2, 0, 3, 4, 5, 0, 4, 5, 5, 1, 4, 8, 0, 0]. Darts_SA的token是：[5, 5, 0, 5, 5, 10, 7, 7, 5, 7, 7, 11, 10, 12, 10, 0, 5, 3, 10, 8].
## 2. 目标检测 ## 2. 目标检测
...@@ -150,20 +149,29 @@ Note: MobileNetV2_NAS 的token是：[4, 4, 5, 1, 1, 2, 1, 1, 0, 2, 6, ...@@ -150,20 +149,29 @@ Note: MobileNetV2_NAS 的token是：[4, 4, 5, 1, 1, 2, 1, 1, 0, 2, 6,
### 2.2 剪裁 ### 2.2 剪裁
数据集：Pasacl VOC & COCO 2017 数据集：Pasacl VOC & COCO 2017
| 模型 | 压缩方法 | 数据集 | Image/GPU | 输入608 Box AP | 输入416 Box AP | 输入320 Box AP | 模型体积(MB) | GFLOPs (608*608) | 下载 | PaddleLite推理耗时说明：
| :----------------------------: | :---------------: | :--------: | :-------: | :------------: | :------------: | :------------: | :----------: | :--------------: | :----------------------------------------------------------: |
| MobileNet-V1-YOLOv3 | Baseline | Pascal VOC | 8 | 76.2 | 76.7 | 75.3 | 94 | 40.49 | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1_voc.tar) | 环境：Qualcomm SnapDragon 845 + armv8
| MobileNet-V1-YOLOv3 | sensitive -52.88% | Pascal VOC | 8 | 77.6 (+1.4) | 77.7 (1.0) | 75.5 (+0.2) | 31 | 19.08 | [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_mobilenet_v1_voc_prune.tar) |
| MobileNet-V1-YOLOv3 | - | COCO | 8 | 29.3 | 29.3 | 27.0 | 95 | 41.35 | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1.tar) | 速度指标：Thread1/Thread2/Thread4耗时
| MobileNet-V1-YOLOv3 | sensitive -51.77% | COCO | 8 | 26.0 (-3.3) | 25.1 (-4.2) | 22.6 (-4.4) | 32 | 19.94 | [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_mobilenet_v1_prune.tar) |
| R50-dcn-YOLOv3 | - | COCO | 8 | 39.1 | - | - | 177 | 89.60 | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_r50vd_dcn.tar) | PaddleLite版本： v2.3
| R50-dcn-YOLOv3 | sensitive -9.37% | COCO | 8 | 39.3 (+0.2) | - | - | 150 | 81.20 | [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_prune.tar) |
| R50-dcn-YOLOv3 | sensitive -24.68% | COCO | 8 | 37.3 (-1.8) | - | - | 113 | 67.48 | [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_prune578.tar) | | 模型 | 压缩方法 | 数据集 | Image/GPU | 输入608 Box AP | 输入416 Box AP | 输入320 Box AP | 模型体积(MB) | GFLOPs (608*608) | PaddleLite推理耗时(ms)(608*608) | TensorRT推理速度(FPS)(608*608) | 下载 |
| R50-dcn-YOLOv3 obj365_pretrain | - | COCO | 8 | 41.4 | - | - | 177 | 89.60 | [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_r50vd_dcn_obj365_pretrained_coco.tar) | | :----------------------------: | :---------------: | :--------: | :-------: | :------------: | :------------: | :------------: | :----------: | :--------------: | :--------------: | :--------------: | :-----------------------------------: |
| R50-dcn-YOLOv3 obj365_pretrain | sensitive -9.37% | COCO | 8 | 40.5 (-0.9) | - | - | 150 | 81.20 | [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_obj365_pretrained_coco_prune.tar) | | MobileNet-V1-YOLOv3 | Baseline | Pascal VOC | 8 | 76.2 | 76.7 | 75.3 | 94 | 40.49 | 1238\796.943\520.101|60.04| [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1_voc.tar) |
| R50-dcn-YOLOv3 obj365_pretrain | sensitive -24.68% | COCO | 8 | 37.8 (-3.3) | - | - | 113 | 67.48 | [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_obj365_pretrained_coco_prune578.tar) | | MobileNet-V1-YOLOv3 | sensitive -52.88% | Pascal VOC | 8 | 77.6 (+1.4) | 77.7 (1.0) | 75.5 (+0.2) | 31 | 19.08 | 602.497\353.759\222.427 |99.36| [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_mobilenet_v1_voc_prune.tar) |
| MobileNet-V1-YOLOv3 | - | COCO | 8 | 29.3 | 29.3 | 27.0 | 95 | 41.35 |-|-| [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v1.tar) |
| MobileNet-V1-YOLOv3 | sensitive -51.77% | COCO | 8 | 26.0 (-3.3) | 25.1 (-4.2) | 22.6 (-4.4) | 32 | 19.94 |-|73.93| [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_mobilenet_v1_prune.tar) |
| R50-dcn-YOLOv3 | - | COCO | 8 | 39.1 | - | - | 177 | 89.60 |-|27.68| [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_r50vd_dcn.tar) |
| R50-dcn-YOLOv3 | sensitive -9.37% | COCO | 8 | 39.3 (+0.2) | - | - | 150 | 81.20 |-|30.08| [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_prune.tar) |
| R50-dcn-YOLOv3 | sensitive -24.68% | COCO | 8 | 37.3 (-1.8) | - | - | 113 | 67.48 |-|34.32| [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_prune578.tar) |
| R50-dcn-YOLOv3 obj365_pretrain | - | COCO | 8 | 41.4 | - | - | 177 | 89.60 |-|-| [下载链接](https://paddlemodels.bj.bcebos.com/object_detection/yolov3_r50vd_dcn_obj365_pretrained_coco.tar) |
| R50-dcn-YOLOv3 obj365_pretrain | sensitive -9.37% | COCO | 8 | 40.5 (-0.9) | - | - | 150 | 81.20 |-|-| [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_obj365_pretrained_coco_prune.tar) |
| R50-dcn-YOLOv3 obj365_pretrain | sensitive -24.68% | COCO | 8 | 37.8 (-3.3) | - | - | 113 | 67.48 |-|-| [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/yolov3_r50vd_dcn_obj365_pretrained_coco_prune578.tar) |
### 2.3 蒸馏 ### 2.3 蒸馏
...@@ -236,8 +244,16 @@ Note: 硬件延时时间是利用提供的硬件延时表得到的，硬件延 ...@@ -236,8 +244,16 @@ Note: 硬件延时时间是利用提供的硬件延时表得到的，硬件延
### 3.2 剪裁 ### 3.2 剪裁
| 模型 | 压缩方法 | mIoU | 模型体积（MB） | GFLOPs | 下载 | PaddleLite推理耗时说明：
| :-------: | :---------------: | :-----------: | :------------: | :----: | :----------------------------------------------------------: |
| fast-scnn | baseline | 69.64 | 11 | 14.41 | [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape.tar) | 环境：Qualcomm SnapDragon 845 + armv8
| fast-scnn | uniform -17.07% | 69.58 (-0.06) | 8.5 | 11.95 | [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape_uniform-17.tar) |
| fast-scnn | sensitive -47.60% | 66.68 (-2.96) | 5.7 | 7.55 | [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape_sensitive-47.tar) | 速度指标：Thread1/Thread2/Thread4耗时
PaddleLite版本： v2.3
| 模型 | 压缩方法 | mIoU | 模型体积（MB） | GFLOPs | PaddleLite推理耗时 | TensorRT推理速度(FPS) | 下载 |
| :-------: | :---------------: | :-----------: | :------------: | :----: | :------------: | :----: | :--------------------------------------: |
| fast-scnn | baseline | 69.64 | 11 | 14.41 | 1226.36\682.96\415.664 |39.53| [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape.tar) |
| fast-scnn | uniform -17.07% | 69.58 (-0.06) | 8.5 | 11.95 | 1140.37\656.612\415.888 |42.01| [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape_uniform-17.tar) |
| fast-scnn | sensitive -47.60% | 66.68 (-2.96) | 5.7 | 7.55 | 866.693\494.467\291.748 |51.48| [下载链接](https://paddlemodels.bj.bcebos.com/PaddleSlim/fast_scnn_cityscape_sensitive-47.tar) |
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