Add Res2Net101_vd and Res2Net200_vd pretrained model (#4180)

PaddleCV/image_classification/README.md

@@ -634,6 +634,8 @@ python -m paddle.distributed.launch train.py \
 |[Res2Net50_26w_4s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net50_26w_4s_pretrained.tar) | 79.33% | 94.57% | 10.731 | 8.274 |
 |[Res2Net50_vd_26w_4s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net50_vd_26w_4s_pretrained.tar) | 79.75% | 94.91% | 11.012 | 8.493 |
 |[Res2Net50_14w_8s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net50_14w_8s_pretrained.tar) | 79.46% | 94.70% | 16.937 | 10.205 |
+|[Res2Net101_vd_26w_4s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net101_vd_26w_4s_pretrained.tar) | 80.64% | 95.22% | 19.612 | 14.651 |
+|[Res2Net200_vd_26w_4s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net200_vd_26w_4s_pretrained.tar) | 81.21% | 95.71% | 35.809 | 26.479 |
 
 ### ResNeXt Series
 |Model | Top-1 | Top-5 | Paddle Fluid inference time(ms) | Paddle TensorRT inference time(ms) |
@@ -789,7 +791,8 @@ python -m paddle.distributed.launch train.py \
 - 2019/09/11 **Stage8**: 更新ResNet18_vd，ResNet34_vd，MobileNetV1_x0_25，MobileNetV1_x0_5，MobileNetV1_x0_75，MobileNetV2_x0_75，MobilenNetV3_small_x1_0，DPN68，DPN92，DPN98，DPN107，DPN131，ResNeXt101_vd_32x4d，ResNeXt152_vd_64x4d，Xception65，Xception71，Xception41_deeplab，Xception65_deeplab，SE_ResNet50_vd
 - 2019/09/20 更新EfficientNet
 - 2019/11/28 **Stage9**: 更新SE_ResNet18_vd，SE_ResNet34_vd，SE_ResNeXt50_vd_32x4d，ResNeXt152_vd_32x4d，Res2Net50_26w_4s，Res2Net50_14w_8s，Res2Net50_vd_26w_4s，HRNet_W18_C，HRNet_W30_C，HRNet_W32_C，HRNet_W40_C，HRNet_W44_C，HRNet_W48_C，HRNet_W64_C
-- 2020/1/7 **Stage10**: 添加AutoDL Series
+- 2020/01/07 **Stage10**: 更新AutoDL Series
+- 2020/01/09 **Stage11**: 更新Res2Net101_vd_26w_4s, Res2Net200_vd_26w_4s
 
 ## 如何贡献代码
 

PaddleCV/image_classification/README_en.md

@@ -471,6 +471,13 @@ Pretrained models can be downloaded by clicking related model names.
 |[ShuffleNetV2_x2_0](https://paddle-imagenet-models-name.bj.bcebos.com/ShuffleNetV2_x2_0_pretrained.tar) | 73.15% | 91.20% | 6.430 | 3.954 |
 |[ShuffleNetV2_swish](https://paddle-imagenet-models-name.bj.bcebos.com/ShuffleNetV2_swish_pretrained.tar) | 70.03% | 89.17% | 6.078 | 4.976 |
 
+### AutoDL Series
+|Model | Top-1 | Top-5 | Paddle Fluid inference time(ms) | Paddle TensorRT inference time(ms) |
+|- |:-: |:-: |:-: |:-: |
+|[DARTS_4M](https://paddle-imagenet-models-name.bj.bcebos.com/DARTS_GS_4M_pretrained.tar) | 75.23% | 92.15% | 13.572 | 6.335 |
+|[DARTS_6M](https://paddle-imagenet-models-name.bj.bcebos.com/DARTS_GS_6M_pretrained.tar) | 76.03% | 92.79% | 16.406 | 6.864 |
+- AutoDL is improved based on DARTS, Local Rademacher Complexity is introduced to control overfitting, and model size is flexibly adjusted through Resource Constraining.
+
 ### ResNet Series
 |Model | Top-1 | Top-5 | Paddle Fluid inference time(ms) | Paddle TensorRT inference time(ms) |
 |- |:-: |:-: |:-: |:-: |
@@ -496,6 +503,8 @@ Pretrained models can be downloaded by clicking related model names.
 |[Res2Net50_26w_4s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net50_26w_4s_pretrained.tar) | 79.33% | 94.57% | 10.731 | 8.274 |
 |[Res2Net50_vd_26w_4s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net50_vd_26w_4s_pretrained.tar) | 79.75% | 94.91% | 11.012 | 8.493 |
 |[Res2Net50_14w_8s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net50_14w_8s_pretrained.tar) | 79.46% | 94.70% | 16.937 | 10.205 |
+|[Res2Net101_vd_26w_4s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net101_vd_26w_4s_pretrained.tar) | 80.64% | 95.22% | 19.612 | 14.651 |
+|[Res2Net200_vd_26w_4s](https://paddle-imagenet-models-name.bj.bcebos.com/Res2Net200_vd_26w_4s_pretrained.tar) | 81.21% | 95.71% | 35.809 | 26.479 |
 
 ### ResNeXt Series
 |Model | Top-1 | Top-5 | Paddle Fluid inference time(ms) | Paddle TensorRT inference time(ms) |
@@ -652,6 +661,8 @@ Enforce failed. Expected x_dims[1] == labels_dims[1], but received x_dims[1]:100
 - 2019/09/11 **Stage8**: Update ResNet18_vd，ResNet34_vd，MobileNetV1_x0_25，MobileNetV1_x0_5，MobileNetV1_x0_75，MobileNetV2_x0_75，MobilenNetV3_small_x1_0，DPN68，DPN92，DPN98，DPN107，DPN131，ResNeXt101_vd_32x4d，ResNeXt152_vd_64x4d，Xception65，Xception71，Xception41_deeplab，Xception65_deeplab，SE_ResNet50_vd
 - 2019/09/20 Update EfficientNet
 - 2019/11/28 **Stage9**: Update SE_ResNet18_vd，SE_ResNet34_vd，SE_ResNeXt50_vd_32x4d，ResNeXt152_vd_32x4d，Res2Net50_26w_4s，Res2Net50_14w_8s，Res2Net50_vd_26w_4s，HRNet_W18_C，HRNet_W30_C，HRNet_W32_C，HRNet_W40_C，HRNet_W44_C，HRNet_W48_C，HRNet_W64_C
+- 2020/01/07 **Stage10**: Update AutoDL Series
+- 2020/01/09 **Stage11**: Update Res2Net101_vd_26w_4s, Res2Net200_vd_26w_4s
 
 ## Contribute
 

PaddleCV/image_classification/scripts/train/Res2Net101_vd_26w_4s.sh

+#Res2Net101_vd_26w_4s
+
+python train.py \
+            --model=Res2Net101_vd_26w_4s \
+            --batch_size=256 \
+            --total_images=1281167 \
+            --class_dim=1000 \
+            --lr_strategy=cosine_decay \
+            --lr=0.1 \
+            --num_epochs=200 \
+            --model_save_dir=output/ \
+            --l2_decay=1e-4 \
+            --use_mixup=True \
+            --use_label_smoothing=True \
+            --label_smoothing_epsilon=0.1

PaddleCV/image_classification/scripts/train/Res2Net200_vd_26w_4s.sh

+#Res2Net200_vd_26w_4s
+
+python train.py \
+            --model=Res2Net200_vd_26w_4s \
+            --batch_size=256 \
+            --total_images=1281167 \
+            --class_dim=1000 \
+            --lr_strategy=cosine_decay \
+            --lr=0.1 \
+            --num_epochs=200 \
+            --model_save_dir=output/ \
+            --l2_decay=1e-4 \
+            --use_mixup=True \
+            --use_label_smoothing=True \
+            --label_smoothing_epsilon=0.1