Skip to content

P
PaddleClas

PaddlePaddle / PaddleClas
大约 2 年 前同步成功

通知 118
Star 4999
Fork 1114
P
PaddleClas
“c339946e6db2ad8ba4c2a91edfa2c63d12908135”上不存在“release/0.10.0/doc/howto/dev/contribute_to_paddle_en.html”
未验证 提交 06cafda7 编写于 作者: W Wei Shengyu 提交者： GitHub
浏览文件
操作

Merge pull request #910 from cuicheng01/develop 

Update ImageNet_models_cn.md and remove expired codes
上级 3579f5a6 b98a1e4e
隐藏空白更改
内联 并排
Showing with 141 addition and 2860 deletion
docs/zh_CN/ImageNet_models_cn.md
浏览文件 @ 06cafda7
......@@ -30,27 +30,26 @@
| 模型 | Top-1 Acc | Reference<br>Top-1 Acc | Acc gain | time(ms)<br>bs=1 | time(ms)<br>bs=4 | Flops(G) | Params(M) | 下载地址 |
|---------------------|-----------|-----------|---------------|----------------|-----------|----------|-----------|-----------------------------------|
| ResNet34_vd_ssld | 0.797 | 0.760 | 0.037 | 2.434 | 6.222 | 7.39 | 21.82 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet34_vd_ssld_pretrained.pdparams) |
| ResNet50_vd_<br>ssld | 0.824 | 0.791 | 0.033 | 3.531 | 8.090 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_ssld_pretrained.pdparams) |
| ResNet50_vd_<br>ssld_v2 | 0.830 | 0.792 | 0.039 | 3.531 | 8.090 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_ssld_v2_pretrained.pdparams) |
| ResNet101_vd_<br>ssld | 0.837 | 0.802 | 0.035 | 6.117 | 13.762 | 16.1 | 44.57 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet101_vd_ssld_pretrained.pdparams) |
| ResNet34_vd_ssld | 0.797 | 0.760 | 0.037 | 2.434 | 6.222 | 7.39 | 21.82 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet34_vd_ssld_pretrained.pdparams) |
| ResNet50_vd_<br>ssld | 0.830 | 0.792 | 0.039 | 3.531 | 8.090 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet50_vd_ssld_pretrained.pdparams) |
| ResNet101_vd_<br>ssld | 0.837 | 0.802 | 0.035 | 6.117 | 13.762 | 16.1 | 44.57 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet101_vd_ssld_pretrained.pdparams) |
| Res2Net50_vd_<br>26w_4s_ssld | 0.831 | 0.798 | 0.033 | 4.527 | 9.657 | 8.37 | 25.06 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/Res2Net50_vd_26w_4s_ssld_pretrained.pdparams) |
| Res2Net101_vd_<br>26w_4s_ssld | 0.839 | 0.806 | 0.033 | 8.087 | 17.312 | 16.67 | 45.22 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/Res2Net101_vd_26w_4s_ssld_pretrained.pdparams) |
| Res2Net200_vd_<br>26w_4s_ssld | 0.851 | 0.812 | 0.049 | 14.678 | 32.350 | 31.49 | 76.21 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/Res2Net200_vd_26w_4s_ssld_pretrained.pdparams) |
| HRNet_W18_C_ssld | 0.812 | 0.769 | 0.043 | 7.406 | 13.297 | 4.14 | 21.29 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W18_C_ssld_pretrained.pdparams) |
| HRNet_W48_C_ssld | 0.836 | 0.790 | 0.046 | 13.707 | 34.435 | 34.58 | 77.47 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W48_C_ssld_pretrained.pdparams) |
| SE_HRNet_W64_C_ssld | 0.848 | - | - | 31.697 | 94.995 | 57.83 | 128.97 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/SE_HRNet_W64_C_ssld_pretrained.pdparams) |
| HRNet_W18_C_ssld | 0.812 | 0.769 | 0.043 | 7.406 | 13.297 | 4.14 | 21.29 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W18_C_ssld_pretrained.pdparams) |
| HRNet_W48_C_ssld | 0.836 | 0.790 | 0.046 | 13.707 | 34.435 | 34.58 | 77.47 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W48_C_ssld_pretrained.pdparams) |
| SE_HRNet_W64_C_ssld | 0.848 | - | - | 31.697 | 94.995 | 57.83 | 128.97 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/SE_HRNet_W64_C_ssld_pretrained.pdparams) |
* 端侧知识蒸馏模型
| 模型 | Top-1 Acc | Reference<br>Top-1 Acc | Acc gain | SD855 time(ms)<br>bs=1 | Flops(G) | Params(M) | 模型大小(M) | 下载地址 |
|---------------------|-----------|-----------|---------------|----------------|-----------|----------|-----------|-----------------------------------|
| MobileNetV1_<br>ssld | 0.779 | 0.710 | 0.069 | 32.523 | 1.11 | 4.19 | 16 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_ssld_pretrained.pdparams) |
| MobileNetV1_<br>ssld | 0.779 | 0.710 | 0.069 | 32.523 | 1.11 | 4.19 | 16 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV1_ssld_pretrained.pdparams) |
| MobileNetV2_<br>ssld | 0.767 | 0.722 | 0.045 | 23.318 | 0.6 | 3.44 | 14 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV2_ssld_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_35_ssld | 0.556 | 0.530 | 0.026 | 2.635 | 0.026 | 1.66 | 6.9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x0_35_ssld_pretrained.pdparams) |
| MobileNetV3_<br>large_x1_0_ssld | 0.790 | 0.753 | 0.036 | 19.308 | 0.45 | 5.47 | 21 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x1_0_ssld_pretrained.pdparams) |
| MobileNetV3_small_<br>x1_0_ssld | 0.713 | 0.682 | 0.031 | 6.546 | 0.123 | 2.94 | 12 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x1_0_ssld_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_35_ssld | 0.556 | 0.530 | 0.026 | 2.635 | 0.026 | 1.66 | 6.9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_small_x0_35_ssld_pretrained.pdparams) |
| MobileNetV3_<br>large_x1_0_ssld | 0.790 | 0.753 | 0.036 | 19.308 | 0.45 | 5.47 | 21 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_large_x1_0_ssld_pretrained.pdparams) |
| MobileNetV3_small_<br>x1_0_ssld | 0.713 | 0.682 | 0.031 | 6.546 | 0.123 | 2.94 | 12 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_small_x1_0_ssld_pretrained.pdparams) |
| GhostNet_<br>x1_3_ssld | 0.794 | 0.757 | 0.037 | 19.983 | 0.44 | 7.3 | 29 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/GhostNet_x1_3_ssld_pretrained.pdparams) |
......@@ -63,23 +62,21 @@ ResNet及其Vd系列模型的精度、速度指标如下表所示，更多关于
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | Flops(G) | Params(M) | 下载地址 |
|---------------------|-----------|-----------|-----------------------|----------------------|----------|-----------|----------------------------------------------------------------------------------------------|
| ResNet18 | 0.7098 | 0.8992 | 1.45606 | 3.56305 | 3.66 | 11.69 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet18_pretrained.pdparams) |
| ResNet18_vd | 0.7226 | 0.9080 | 1.54557 | 3.85363 | 4.14 | 11.71 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet18_vd_pretrained.pdparams) |
| ResNet34 | 0.7457 | 0.9214 | 2.34957 | 5.89821 | 7.36 | 21.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet34_pretrained.pdparams) |
| ResNet34_vd | 0.7598 | 0.9298 | 2.43427 | 6.22257 | 7.39 | 21.82 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet34_vd_pretrained.pdparams) |
| ResNet34_vd_ssld | 0.7972 | 0.9490 | 2.43427 | 6.22257 | 7.39 | 21.82 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet34_vd_ssld_pretrained.pdparams) |
| ResNet50 | 0.7650 | 0.9300 | 3.47712 | 7.84421 | 8.19 | 25.56 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_pretrained.pdparams) |
| ResNet18 | 0.7098 | 0.8992 | 1.45606 | 3.56305 | 3.66 | 11.69 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet18_pretrained.pdparams) |
| ResNet18_vd | 0.7226 | 0.9080 | 1.54557 | 3.85363 | 4.14 | 11.71 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet18_vd_pretrained.pdparams) |
| ResNet34 | 0.7457 | 0.9214 | 2.34957 | 5.89821 | 7.36 | 21.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet34_pretrained.pdparams) |
| ResNet34_vd | 0.7598 | 0.9298 | 2.43427 | 6.22257 | 7.39 | 21.82 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet34_vd_pretrained.pdparams) |
| ResNet34_vd_ssld | 0.7972 | 0.9490 | 2.43427 | 6.22257 | 7.39 | 21.82 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet34_vd_ssld_pretrained.pdparams) |
| ResNet50 | 0.7650 | 0.9300 | 3.47712 | 7.84421 | 8.19 | 25.56 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet50_pretrained.pdparams) |
| ResNet50_vc | 0.7835 | 0.9403 | 3.52346 | 8.10725 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vc_pretrained.pdparams) |
| ResNet50_vd | 0.7912 | 0.9444 | 3.53131 | 8.09057 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_pretrained.pdparams) |
| ResNet50_vd_v2 | 0.7984 | 0.9493 | 3.53131 | 8.09057 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_v2_pretrained.pdparams) |
| ResNet101 | 0.7756 | 0.9364 | 6.07125 | 13.40573 | 15.52 | 44.55 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet101_pretrained.pdparams) |
| ResNet101_vd | 0.8017 | 0.9497 | 6.11704 | 13.76222 | 16.1 | 44.57 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet101_vd_pretrained.pdparams) |
| ResNet152 | 0.7826 | 0.9396 | 8.50198 | 19.17073 | 23.05 | 60.19 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet152_pretrained.pdparams) |
| ResNet152_vd | 0.8059 | 0.9530 | 8.54376 | 19.52157 | 23.53 | 60.21 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet152_vd_pretrained.pdparams) |
| ResNet200_vd | 0.8093 | 0.9533 | 10.80619 | 25.01731 | 30.53 | 74.74 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet200_vd_pretrained.pdparams) |
| ResNet50_vd_<br>ssld | 0.8239 | 0.9610 | 3.53131 | 8.09057 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_ssld_pretrained.pdparams) |
| ResNet50_vd_<br>ssld_v2 | 0.8300 | 0.9640 | 3.53131 | 8.09057 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_ssld_v2_pretrained.pdparams) |
| ResNet101_vd_<br>ssld | 0.8373 | 0.9669 | 6.11704 | 13.76222 | 16.1 | 44.57 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet101_vd_ssld_pretrained.pdparams) |
| ResNet50_vd | 0.7912 | 0.9444 | 3.53131 | 8.09057 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet50_vd_pretrained.pdparams) |
| ResNet101 | 0.7756 | 0.9364 | 6.07125 | 13.40573 | 15.52 | 44.55 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet101_pretrained.pdparams) |
| ResNet101_vd | 0.8017 | 0.9497 | 6.11704 | 13.76222 | 16.1 | 44.57 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet101_vd_pretrained.pdparams) |
| ResNet152 | 0.7826 | 0.9396 | 8.50198 | 19.17073 | 23.05 | 60.19 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet152_pretrained.pdparams) |
| ResNet152_vd | 0.8059 | 0.9530 | 8.54376 | 19.52157 | 23.53 | 60.21 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet152_vd_pretrained.pdparams) |
| ResNet200_vd | 0.8093 | 0.9533 | 10.80619 | 25.01731 | 30.53 | 74.74 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet200_vd_pretrained.pdparams) |
| ResNet50_vd_<br>ssld | 0.8300 | 0.9640 | 3.53131 | 8.09057 | 8.67 | 25.58 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet50_vd_ssld_pretrained.pdparams) |
| ResNet101_vd_<br>ssld | 0.8373 | 0.9669 | 6.11704 | 13.76222 | 16.1 | 44.57 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ResNet101_vd_ssld_pretrained.pdparams) |
<a name="移动端系列"></a>
......@@ -89,11 +86,11 @@ ResNet及其Vd系列模型的精度、速度指标如下表所示，更多关于
| 模型 | Top-1 Acc | Top-5 Acc | SD855 time(ms)<br>bs=1 | Flops(G) | Params(M) | 模型大小(M) | 下载地址 |
|----------------------------------|-----------|-----------|------------------------|----------|-----------|---------|-----------------------------------------------------------------------------------------------------------|
| MobileNetV1_<br>x0_25 | 0.5143 | 0.7546 | 3.21985 | 0.07 | 0.46 | 1.9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_x0_25_pretrained.pdparams) |
| MobileNetV1_<br>x0_5 | 0.6352 | 0.8473 | 9.579599 | 0.28 | 1.31 | 5.2 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_x0_5_pretrained.pdparams) |
| MobileNetV1_<br>x0_75 | 0.6881 | 0.8823 | 19.436399 | 0.63 | 2.55 | 10 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_x0_75_pretrained.pdparams) |
| MobileNetV1 | 0.7099 | 0.8968 | 32.523048 | 1.11 | 4.19 | 16 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_pretrained.pdparams) |
| MobileNetV1_<br>ssld | 0.7789 | 0.9394 | 32.523048 | 1.11 | 4.19 | 16 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_ssld_pretrained.pdparams) |
| MobileNetV1_<br>x0_25 | 0.5143 | 0.7546 | 3.21985 | 0.07 | 0.46 | 1.9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV1_x0_25_pretrained.pdparams) |
| MobileNetV1_<br>x0_5 | 0.6352 | 0.8473 | 9.579599 | 0.28 | 1.31 | 5.2 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV1_x0_5_pretrained.pdparams) |
| MobileNetV1_<br>x0_75 | 0.6881 | 0.8823 | 19.436399 | 0.63 | 2.55 | 10 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV1_x0_75_pretrained.pdparams) |
| MobileNetV1 | 0.7099 | 0.8968 | 32.523048 | 1.11 | 4.19 | 16 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV1_pretrained.pdparams) |
| MobileNetV1_<br>ssld | 0.7789 | 0.9394 | 32.523048 | 1.11 | 4.19 | 16 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV1_ssld_pretrained.pdparams) |
| MobileNetV2_<br>x0_25 | 0.5321 | 0.7652 | 3.79925 | 0.05 | 1.5 | 6.1 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV2_x0_25_pretrained.pdparams) |
| MobileNetV2_<br>x0_5 | 0.6503 | 0.8572 | 8.7021 | 0.17 | 1.93 | 7.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV2_x0_5_pretrained.pdparams) |
| MobileNetV2_<br>x0_75 | 0.6983 | 0.8901 | 15.531351 | 0.35 | 2.58 | 10 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV2_x0_75_pretrained.pdparams) |
......@@ -101,19 +98,19 @@ ResNet及其Vd系列模型的精度、速度指标如下表所示，更多关于
| MobileNetV2_<br>x1_5 | 0.7412 | 0.9167 | 45.623848 | 1.32 | 6.76 | 26 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV2_x1_5_pretrained.pdparams) |
| MobileNetV2_<br>x2_0 | 0.7523 | 0.9258 | 74.291649 | 2.32 | 11.13 | 43 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV2_x2_0_pretrained.pdparams) |
| MobileNetV2_<br>ssld | 0.7674 | 0.9339 | 23.317699 | 0.6 | 3.44 | 14 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV2_ssld_pretrained.pdparams) |
| MobileNetV3_<br>large_x1_25 | 0.7641 | 0.9295 | 28.217701 | 0.714 | 7.44 | 29 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x1_25_pretrained.pdparams) |
| MobileNetV3_<br>large_x1_0 | 0.7532 | 0.9231 | 19.30835 | 0.45 | 5.47 | 21 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x1_0_pretrained.pdparams) |
| MobileNetV3_<br>large_x0_75 | 0.7314 | 0.9108 | 13.5646 | 0.296 | 3.91 | 16 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x0_75_pretrained.pdparams) |
| MobileNetV3_<br>large_x0_5 | 0.6924 | 0.8852 | 7.49315 | 0.138 | 2.67 | 11 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x0_5_pretrained.pdparams) |
| MobileNetV3_<br>large_x0_35 | 0.6432 | 0.8546 | 5.13695 | 0.077 | 2.1 | 8.6 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x0_35_pretrained.pdparams) |
| MobileNetV3_<br>small_x1_25 | 0.7067 | 0.8951 | 9.2745 | 0.195 | 3.62 | 14 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x1_25_pretrained.pdparams) |
| MobileNetV3_<br>small_x1_0 | 0.6824 | 0.8806 | 6.5463 | 0.123 | 2.94 | 12 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x1_0_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_75 | 0.6602 | 0.8633 | 5.28435 | 0.088 | 2.37 | 9.6 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x0_75_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_5 | 0.5921 | 0.8152 | 3.35165 | 0.043 | 1.9 | 7.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x0_5_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_35 | 0.5303 | 0.7637 | 2.6352 | 0.026 | 1.66 | 6.9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x0_35_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_35_ssld | 0.5555 | 0.7771 | 2.6352 | 0.026 | 1.66 | 6.9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x0_35_ssld_pretrained.pdparams) |
| MobileNetV3_<br>large_x1_0_ssld | 0.7896 | 0.9448 | 19.30835 | 0.45 | 5.47 | 21 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x1_0_ssld_pretrained.pdparams) |
| MobileNetV3_small_<br>x1_0_ssld | 0.7129 | 0.9010 | 6.5463 | 0.123 | 2.94 | 12 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x1_0_ssld_pretrained.pdparams) |
| MobileNetV3_<br>large_x1_25 | 0.7641 | 0.9295 | 28.217701 | 0.714 | 7.44 | 29 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_large_x1_25_pretrained.pdparams) |
| MobileNetV3_<br>large_x1_0 | 0.7532 | 0.9231 | 19.30835 | 0.45 | 5.47 | 21 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_large_x1_0_pretrained.pdparams) |
| MobileNetV3_<br>large_x0_75 | 0.7314 | 0.9108 | 13.5646 | 0.296 | 3.91 | 16 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_large_x0_75_pretrained.pdparams) |
| MobileNetV3_<br>large_x0_5 | 0.6924 | 0.8852 | 7.49315 | 0.138 | 2.67 | 11 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_large_x0_5_pretrained.pdparams) |
| MobileNetV3_<br>large_x0_35 | 0.6432 | 0.8546 | 5.13695 | 0.077 | 2.1 | 8.6 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_large_x0_35_pretrained.pdparams) |
| MobileNetV3_<br>small_x1_25 | 0.7067 | 0.8951 | 9.2745 | 0.195 | 3.62 | 14 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_small_x1_25_pretrained.pdparams) |
| MobileNetV3_<br>small_x1_0 | 0.6824 | 0.8806 | 6.5463 | 0.123 | 2.94 | 12 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_small_x1_0_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_75 | 0.6602 | 0.8633 | 5.28435 | 0.088 | 2.37 | 9.6 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_small_x0_75_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_5 | 0.5921 | 0.8152 | 3.35165 | 0.043 | 1.9 | 7.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_small_x0_5_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_35 | 0.5303 | 0.7637 | 2.6352 | 0.026 | 1.66 | 6.9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_small_x0_35_pretrained.pdparams) |
| MobileNetV3_<br>small_x0_35_ssld | 0.5555 | 0.7771 | 2.6352 | 0.026 | 1.66 | 6.9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_small_x0_35_ssld_pretrained.pdparams) |
| MobileNetV3_<br>large_x1_0_ssld | 0.7896 | 0.9448 | 19.30835 | 0.45 | 5.47 | 21 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_large_x1_0_ssld_pretrained.pdparams) |
| MobileNetV3_small_<br>x1_0_ssld | 0.7129 | 0.9010 | 6.5463 | 0.123 | 2.94 | 12 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/MobileNetV3_small_x1_0_ssld_pretrained.pdparams) |
| ShuffleNetV2 | 0.6880 | 0.8845 | 10.941 | 0.28 | 2.26 | 9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ShuffleNetV2_x1_0_pretrained.pdparams) |
| ShuffleNetV2_<br>x0_25 | 0.4990 | 0.7379 | 2.329 | 0.03 | 0.6 | 2.7 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ShuffleNetV2_x0_25_pretrained.pdparams) |
| ShuffleNetV2_<br>x0_33 | 0.5373 | 0.7705 | 2.64335 | 0.04 | 0.64 | 2.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ShuffleNetV2_x0_33_pretrained.pdparams) |
......@@ -191,16 +188,16 @@ HRNet系列模型的精度、速度指标如下表所示，更多关于该系列
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | Flops(G) | Params(M) | 下载地址 |
|-------------|-----------|-----------|------------------|------------------|----------|-----------|--------------------------------------------------------------------------------------|
| HRNet_W18_C | 0.7692 | 0.9339 | 7.40636 | 13.29752 | 4.14 | 21.29 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W18_C_pretrained.pdparams) |
| HRNet_W18_C_ssld | 0.81162 | 0.95804 | 7.40636 | 13.29752 | 4.14 | 21.29 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W18_C_ssld_pretrained.pdparams) |
| HRNet_W30_C | 0.7804 | 0.9402 | 9.57594 | 17.35485 | 16.23 | 37.71 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W30_C_pretrained.pdparams) |
| HRNet_W32_C | 0.7828 | 0.9424 | 9.49807 | 17.72921 | 17.86 | 41.23 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W32_C_pretrained.pdparams) |
| HRNet_W40_C | 0.7877 | 0.9447 | 12.12202 | 25.68184 | 25.41 | 57.55 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W40_C_pretrained.pdparams) |
| HRNet_W44_C | 0.7900 | 0.9451 | 13.19858 | 32.25202 | 29.79 | 67.06 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W44_C_pretrained.pdparams) |
| HRNet_W48_C | 0.7895 | 0.9442 | 13.70761 | 34.43572 | 34.58 | 77.47 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W48_C_pretrained.pdparams) |
| HRNet_W48_C_ssld | 0.8363 | 0.9682 | 13.70761 | 34.43572 | 34.58 | 77.47 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W48_C_ssld_pretrained.pdparams) |
| HRNet_W64_C | 0.7930 | 0.9461 | 17.57527 | 47.9533 | 57.83 | 128.06 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W64_C_pretrained.pdparams) |
| SE_HRNet_W64_C_ssld | 0.8475 | 0.9726 | 31.69770 | 94.99546 | 57.83 | 128.97 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/SE_HRNet_W64_C_ssld_pretrained.pdparams) |
| HRNet_W18_C | 0.7692 | 0.9339 | 7.40636 | 13.29752 | 4.14 | 21.29 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W18_C_pretrained.pdparams) |
| HRNet_W18_C_ssld | 0.81162 | 0.95804 | 7.40636 | 13.29752 | 4.14 | 21.29 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W18_C_ssld_pretrained.pdparams) |
| HRNet_W30_C | 0.7804 | 0.9402 | 9.57594 | 17.35485 | 16.23 | 37.71 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W30_C_pretrained.pdparams) |
| HRNet_W32_C | 0.7828 | 0.9424 | 9.49807 | 17.72921 | 17.86 | 41.23 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W32_C_pretrained.pdparams) |
| HRNet_W40_C | 0.7877 | 0.9447 | 12.12202 | 25.68184 | 25.41 | 57.55 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W40_C_pretrained.pdparams) |
| HRNet_W44_C | 0.7900 | 0.9451 | 13.19858 | 32.25202 | 29.79 | 67.06 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W44_C_pretrained.pdparams) |
| HRNet_W48_C | 0.7895 | 0.9442 | 13.70761 | 34.43572 | 34.58 | 77.47 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W48_C_pretrained.pdparams) |
| HRNet_W48_C_ssld | 0.8363 | 0.9682 | 13.70761 | 34.43572 | 34.58 | 77.47 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W48_C_ssld_pretrained.pdparams) |
| HRNet_W64_C | 0.7930 | 0.9461 | 17.57527 | 47.9533 | 57.83 | 128.06 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/HRNet_W64_C_pretrained.pdparams) |
| SE_HRNet_W64_C_ssld | 0.8475 | 0.9726 | 31.69770 | 94.99546 | 57.83 | 128.97 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/SE_HRNet_W64_C_ssld_pretrained.pdparams) |
<a name="Inception系列"></a>
......@@ -216,7 +213,7 @@ Inception系列模型的精度、速度指标如下表所示，更多关于该
| Xception65 | 0.8100 | 0.9549 | 7.26158 | 25.88778 | 25.95 | 35.48 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/Xception65_pretrained.pdparams) |
| Xception65_deeplab | 0.8032 | 0.9449 | 7.60208 | 26.03699 | 27.37 | 39.52 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/Xception65_deeplab_pretrained.pdparams) |
| Xception71 | 0.8111 | 0.9545 | 8.72457 | 31.55549 | 31.77 | 37.28 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/Xception71_pretrained.pdparams) |
| InceptionV3 | 0.7914 | 0.9459 | 6.64054 | 13.53630 | 11.46 | 23.83 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/InceptionV3_pretrained.pdparams) |
| InceptionV3 | 0.7914 | 0.9459 | 6.64054 | 13.53630 | 11.46 | 23.83 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/InceptionV3_pretrained.pdparams) |
| InceptionV4 | 0.8077 | 0.9526 | 12.99342 | 25.23416 | 24.57 | 42.68 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/InceptionV4_pretrained.pdparams) |
......@@ -352,6 +349,90 @@ ViT（Vision Transformer）与DeiT（Data-efficient Image Transformers）系列
[1]：基于ImageNet22k数据集预训练，然后在ImageNet1k数据集迁移学习得到。
<a name="LeViT系列"></a>
### LeViT系列
关于LeViT系列模型的精度、速度指标如下表所示，更多介绍可以参考：[LeViT系列模型文档](./models/LeViT.md)。
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | Flops(M) | Params(M) | 下载地址 |
| ---------- | --------- | --------- | ---------------- | ---------------- | -------- | --------- | ------------------------------------------------------------ |
| LeViT_128S | 0.7598 | 0.9269 | | | 305 | 7.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/LeViT_128S_pretrained.pdparams) |
| LeViT_128 | 0.7810 | 0.9371 | | | 406 | 9.2 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/LeViT_128_pretrained.pdparams) |
| LeViT_192 | 0.7934 | 0.9446 | | | 658 | 11 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/LeViT_192_pretrained.pdparams) |
| LeViT_256 | 0.8085 | 0.9497 | | | 1120 | 19 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/LeViT_256_pretrained.pdparams) |
| LeViT_384 | 0.8191 | 0.9551 | | | 2353 | 39 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/LeViT_384_pretrained.pdparams) |
**注**：与Reference的精度差异源于数据预处理不同及未使用蒸馏的head作为输出。
<a name="Twins系列"></a>
### Twins系列
关于Twins系列模型的精度、速度指标如下表所示，更多介绍可以参考：[Twins系列模型文档](./models/Twins.md)。
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | Flops(G) | Params(M) | 下载地址 |
| ---------- | --------- | --------- | ---------------- | ---------------- | -------- | --------- | ------------------------------------------------------------ |
| pcpvt_small | 0.8082 | 0.9552 | | |3.7 | 24.1 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/pcpvt_small_pretrained.pdparams) |
| pcpvt_base | 0.8242 | 0.9619 | | | 6.4 | 43.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/pcpvt_base_pretrained.pdparams) |
| pcpvt_large | 0.8273 | 0.9650 | | | 9.5 | 60.9 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/pcpvt_large_pretrained.pdparams) |
| alt_gvt_small | 0.8140 | 0.9546 | | |2.8 | 24 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/alt_gvt_small_pretrained.pdparams) |
| alt_gvt_base | 0.8294 | 0.9621 | | | 8.3 | 56 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/alt_gvt_base_pretrained.pdparams) |
| alt_gvt_large | 0.8331 | 0.9642 | | | 14.8 | 99.2 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/alt_gvt_large_pretrained.pdparams) |
**注**：与Reference的精度差异源于数据预处理不同。
<a name="HarDNet系列"></a>
### HarDNet系列
关于HarDNet系列模型的精度、速度指标如下表所示，更多介绍可以参考：[HarDNet系列模型文档](./models/HarDNet.md)。
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | Flops(G) | Params(M) | 下载地址 |
| ---------- | --------- | --------- | ---------------- | ---------------- | -------- | --------- | ------------------------------------------------------------ |
| HarDNet39_ds | 0.7133 |0.8998 | | | 0.4 | 3.5 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HarDNet39_ds_pretrained.pdparams) |
| HarDNet68_ds |0.7362 | 0.9152 | | | 0.8 | 4.2 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HarDNet68_ds_pretrained.pdparams) |
| HarDNet68| 0.7546 | 0.9265 | | | 4.3 | 17.6 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HarDNet68_pretrained.pdparams) |
| HarDNet85 | 0.7744 | 0.9355 | | | 9.1 | 36.7 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HarDNet85_pretrained.pdparams) |
<a name="DLA系列"></a>
### DLA系列
关于 DLA系列模型的精度、速度指标如下表所示，更多介绍可以参考：[DLA系列模型文档](./models/DLA.md)。
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | Flops(G) | Params(M) | 下载地址 |
| ---------- | --------- | --------- | ---------------- | ---------------- | -------- | --------- | ------------------------------------------------------------ |
| DLA102 | 0.7893 |0.9452 | | | 7.2 | 33.3 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DLA102_pretrained.pdparams) |
| DLA102x2 |0.7885 | 0.9445 | | | 9.3 | 41.4 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DLA102x2_pretrained.pdparams) |
| DLA102x| 0.781 | 0.9400 | | | 5.9 | 26.4 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DLA102x_pretrained.pdparams) |
| DLA169 | 0.7809 | 0.9409 | | | 11.6 | 53.5 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DLA169_pretrained.pdparams) |
| DLA34 | 0.7603 | 0.9298 | | | 3.1 | 15.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DLA34_pretrained.pdparams) |
| DLA46_c |0.6321 | 0.853 | | | 0.5 | 1.3 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DLA46_c_pretrained.pdparams) |
| DLA60 | 0.7610 | 0.9292 | | | 4.2 | 22.0 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DLA60_pretrained.pdparams) |
| DLA60x_c | 0.6645 | 0.8754 | | | 0.6 | 1.3 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DLA60x_c_pretrained.pdparams) |
| DLA60x | 0.7753 | 0.9378 | | | 3.5 | 17.4 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DLA60x_pretrained.pdparams) |
<a name="RedNet系列"></a>
### RedNet系列
关于RedNet系列模型的精度、速度指标如下表所示，更多介绍可以参考：[RedNet系列模型文档](./models/RedNet.md)。
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | Flops(G) | Params(M) | 下载地址 |
| ---------- | --------- | --------- | ---------------- | ---------------- | -------- | --------- | ------------------------------------------------------------ |
| RedNet26 | 0.7595 |0.9319 | | | 1.7 | 9.2 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RedNet26_pretrained.pdparams) |
| RedNet38 |0.7747 | 0.9356 | | | 2.2 | 12.4 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RedNet38_pretrained.pdparams) |
| RedNet50| 0.7833 | 0.9417 | | | 2.7 | 15.5 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RedNet50_pretrained.pdparams) |
| RedNet101 | 0.7894 | 0.9436 | | | 4.7 | 25.7 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RedNet101_pretrained.pdparams) |
| RedNet152 | 0.7917 | 0.9440 | | | 6.8 | 34.0 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/RedNet152_pretrained.pdparams) |
<a name="TNT系列"></a>
### TNT系列
关于TNT系列模型的精度、速度指标如下表所示，更多介绍可以参考：[TNT系列模型文档](./models/TNT.md)。
| 模型 | Top-1 Acc | Top-5 Acc | time(ms)<br>bs=1 | time(ms)<br>bs=4 | Flops(G) | Params(M) | 下载地址 |
| ---------- | --------- | --------- | ---------------- | ---------------- | -------- | --------- | ------------------------------------------------------------ |
| TNT_small | 0.8121 |0.9563 | | | 5.2 | 23.8 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/TNT_small_pretrained.pdparams) | |
**注**：TNT模型的数据预处理部分`NormalizeImage`中的`mean`与`std`均为0.5。
<a name="其他模型"></a>
### 其他模型
......@@ -364,8 +445,8 @@ ViT（Vision Transformer）与DeiT（Data-efficient Image Transformers）系列
| AlexNet | 0.567 | 0.792 | 1.44993 | 2.46696 | 1.370 | 61.090 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/AlexNet_pretrained.pdparams) |
| SqueezeNet1_0 | 0.596 | 0.817 | 0.96736 | 2.53221 | 1.550 | 1.240 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/SqueezeNet1_0_pretrained.pdparams) |
| SqueezeNet1_1 | 0.601 | 0.819 | 0.76032 | 1.877 | 0.690 | 1.230 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/SqueezeNet1_1_pretrained.pdparams) |
| VGG11 | 0.693 | 0.891 | 3.90412 | 9.51147 | 15.090 | 132.850 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/VGG11_pretrained.pdparams) |
| VGG13 | 0.700 | 0.894 | 4.64684 | 12.61558 | 22.480 | 133.030 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/VGG13_pretrained.pdparams) |
| VGG16 | 0.720 | 0.907 | 5.61769 | 16.40064 | 30.810 | 138.340 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/VGG16_pretrained.pdparams) |
| VGG19 | 0.726 | 0.909 | 6.65221 | 20.4334 | 39.130 | 143.650 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/VGG19_pretrained.pdparams) |
| VGG11 | 0.693 | 0.891 | 3.90412 | 9.51147 | 15.090 | 132.850 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/VGG11_pretrained.pdparams) |
| VGG13 | 0.700 | 0.894 | 4.64684 | 12.61558 | 22.480 | 133.030 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/VGG13_pretrained.pdparams) |
| VGG16 | 0.720 | 0.907 | 5.61769 | 16.40064 | 30.810 | 138.340 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/VGG16_pretrained.pdparams) |
| VGG19 | 0.726 | 0.909 | 6.65221 | 20.4334 | 39.130 | 143.650 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/VGG19_pretrained.pdparams) |
| DarkNet53 | 0.780 | 0.941 | 4.10829 | 12.1714 | 18.580 | 41.600 | [下载链接](https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/DarkNet53_pretrained.pdparams) |
ppcls/arch/backbone/model_zoo/hrnet.py 已删除 100644 → 0
浏览文件 @ 3579f5a6
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import numpy as np
import paddle
from paddle import ParamAttr
import paddle.nn as nn
import paddle.nn.functional as F
from paddle.nn import Conv2D, BatchNorm, Linear
from paddle.nn import AdaptiveAvgPool2D, MaxPool2D, AvgPool2D
from paddle.nn.initializer import Uniform
import math
from ppcls.utils.save_load import load_dygraph_pretrain, load_dygraph_pretrain_from_url
MODEL_URLS = {"HRNet_W18_C": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W18_C_pretrained.pdparams",
"HRNet_W30_C": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W30_C_pretrained.pdparams",
"HRNet_W32_C": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W32_C_pretrained.pdparams",
"HRNet_W40_C": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W40_C_pretrained.pdparams",
"HRNet_W44_C": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W44_C_pretrained.pdparams",
"HRNet_W48_C": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W48_C_pretrained.pdparams",
"HRNet_W64_C": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/HRNet_W64_C_pretrained.pdparams",
}
__all__ = list(MODEL_URLS.keys())
class ConvBNLayer(nn.Layer):
def __init__(self,
num_channels,
num_filters,
filter_size,
stride=1,
groups=1,
act="relu",
name=None):
super(ConvBNLayer, self).__init__()
self._conv = Conv2D(
in_channels=num_channels,
out_channels=num_filters,
kernel_size=filter_size,
stride=stride,
padding=(filter_size - 1) // 2,
groups=groups,
weight_attr=ParamAttr(name=name + "_weights"),
bias_attr=False)
bn_name = name + '_bn'
self._batch_norm = BatchNorm(
num_filters,
act=act,
param_attr=ParamAttr(name=bn_name + '_scale'),
bias_attr=ParamAttr(bn_name + '_offset'),
moving_mean_name=bn_name + '_mean',
moving_variance_name=bn_name + '_variance')
def forward(self, input):
y = self._conv(input)
y = self._batch_norm(y)
return y
class Layer1(nn.Layer):
def __init__(self, num_channels, has_se=False, name=None):
super(Layer1, self).__init__()
self.bottleneck_block_list = []
for i in range(4):
bottleneck_block = self.add_sublayer(
"bb_{}_{}".format(name, i + 1),
BottleneckBlock(
num_channels=num_channels if i == 0 else 256,
num_filters=64,
has_se=has_se,
stride=1,
downsample=True if i == 0 else False,
name=name + '_' + str(i + 1)))
self.bottleneck_block_list.append(bottleneck_block)
def forward(self, input):
conv = input
for block_func in self.bottleneck_block_list:
conv = block_func(conv)
return conv
class TransitionLayer(nn.Layer):
def __init__(self, in_channels, out_channels, name=None):
super(TransitionLayer, self).__init__()
num_in = len(in_channels)
num_out = len(out_channels)
out = []
self.conv_bn_func_list = []
for i in range(num_out):
residual = None
if i < num_in:
if in_channels[i] != out_channels[i]:
residual = self.add_sublayer(
"transition_{}_layer_{}".format(name, i + 1),
ConvBNLayer(
num_channels=in_channels[i],
num_filters=out_channels[i],
filter_size=3,
name=name + '_layer_' + str(i + 1)))
else:
residual = self.add_sublayer(
"transition_{}_layer_{}".format(name, i + 1),
ConvBNLayer(
num_channels=in_channels[-1],
num_filters=out_channels[i],
filter_size=3,
stride=2,
name=name + '_layer_' + str(i + 1)))
self.conv_bn_func_list.append(residual)
def forward(self, input):
outs = []
for idx, conv_bn_func in enumerate(self.conv_bn_func_list):
if conv_bn_func is None:
outs.append(input[idx])
else:
if idx < len(input):
outs.append(conv_bn_func(input[idx]))
else:
outs.append(conv_bn_func(input[-1]))
return outs
class Branches(nn.Layer):
def __init__(self,
block_num,
in_channels,
out_channels,
has_se=False,
name=None):
super(Branches, self).__init__()
self.basic_block_list = []
for i in range(len(out_channels)):
self.basic_block_list.append([])
for j in range(block_num):
in_ch = in_channels[i] if j == 0 else out_channels[i]
basic_block_func = self.add_sublayer(
"bb_{}_branch_layer_{}_{}".format(name, i + 1, j + 1),
BasicBlock(
num_channels=in_ch,
num_filters=out_channels[i],
has_se=has_se,
name=name + '_branch_layer_' + str(i + 1) + '_' +
str(j + 1)))
self.basic_block_list[i].append(basic_block_func)
def forward(self, inputs):
outs = []
for idx, input in enumerate(inputs):
conv = input
basic_block_list = self.basic_block_list[idx]
for basic_block_func in basic_block_list:
conv = basic_block_func(conv)
outs.append(conv)
return outs
class BottleneckBlock(nn.Layer):
def __init__(self,
num_channels,
num_filters,
has_se,
stride=1,
downsample=False,
name=None):
super(BottleneckBlock, self).__init__()
self.has_se = has_se
self.downsample = downsample
self.conv1 = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters,
filter_size=1,
act="relu",
name=name + "_conv1", )
self.conv2 = ConvBNLayer(
num_channels=num_filters,
num_filters=num_filters,
filter_size=3,
stride=stride,
act="relu",
name=name + "_conv2")
self.conv3 = ConvBNLayer(
num_channels=num_filters,
num_filters=num_filters * 4,
filter_size=1,
act=None,
name=name + "_conv3")
if self.downsample:
self.conv_down = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters * 4,
filter_size=1,
act=None,
name=name + "_downsample")
if self.has_se:
self.se = SELayer(
num_channels=num_filters * 4,
num_filters=num_filters * 4,
reduction_ratio=16,
name='fc' + name)
def forward(self, input):
residual = input
conv1 = self.conv1(input)
conv2 = self.conv2(conv1)
conv3 = self.conv3(conv2)
if self.downsample:
residual = self.conv_down(input)
if self.has_se:
conv3 = self.se(conv3)
y = paddle.add(x=residual, y=conv3)
y = F.relu(y)
return y
class BasicBlock(nn.Layer):
def __init__(self,
num_channels,
num_filters,
stride=1,
has_se=False,
downsample=False,
name=None):
super(BasicBlock, self).__init__()
self.has_se = has_se
self.downsample = downsample
self.conv1 = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters,
filter_size=3,
stride=stride,
act="relu",
name=name + "_conv1")
self.conv2 = ConvBNLayer(
num_channels=num_filters,
num_filters=num_filters,
filter_size=3,
stride=1,
act=None,
name=name + "_conv2")
if self.downsample:
self.conv_down = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters * 4,
filter_size=1,
act="relu",
name=name + "_downsample")
if self.has_se:
self.se = SELayer(
num_channels=num_filters,
num_filters=num_filters,
reduction_ratio=16,
name='fc' + name)
def forward(self, input):
residual = input
conv1 = self.conv1(input)
conv2 = self.conv2(conv1)
if self.downsample:
residual = self.conv_down(input)
if self.has_se:
conv2 = self.se(conv2)
y = paddle.add(x=residual, y=conv2)
y = F.relu(y)
return y
class SELayer(nn.Layer):
def __init__(self, num_channels, num_filters, reduction_ratio, name=None):
super(SELayer, self).__init__()
self.pool2d_gap = AdaptiveAvgPool2D(1)
self._num_channels = num_channels
med_ch = int(num_channels / reduction_ratio)
stdv = 1.0 / math.sqrt(num_channels * 1.0)
self.squeeze = Linear(
num_channels,
med_ch,
weight_attr=ParamAttr(
initializer=Uniform(-stdv, stdv), name=name + "_sqz_weights"),
bias_attr=ParamAttr(name=name + '_sqz_offset'))
stdv = 1.0 / math.sqrt(med_ch * 1.0)
self.excitation = Linear(
med_ch,
num_filters,
weight_attr=ParamAttr(
initializer=Uniform(-stdv, stdv), name=name + "_exc_weights"),
bias_attr=ParamAttr(name=name + '_exc_offset'))
def forward(self, input):
pool = self.pool2d_gap(input)
pool = paddle.squeeze(pool, axis=[2, 3])
squeeze = self.squeeze(pool)
squeeze = F.relu(squeeze)
excitation = self.excitation(squeeze)
excitation = F.sigmoid(excitation)
excitation = paddle.unsqueeze(excitation, axis=[2, 3])
out = input * excitation
return out
class Stage(nn.Layer):
def __init__(self,
num_channels,
num_modules,
num_filters,
has_se=False,
multi_scale_output=True,
name=None):
super(Stage, self).__init__()
self._num_modules = num_modules
self.stage_func_list = []
for i in range(num_modules):
if i == num_modules - 1 and not multi_scale_output:
stage_func = self.add_sublayer(
"stage_{}_{}".format(name, i + 1),
HighResolutionModule(
num_channels=num_channels,
num_filters=num_filters,
has_se=has_se,
multi_scale_output=False,
name=name + '_' + str(i + 1)))
else:
stage_func = self.add_sublayer(
"stage_{}_{}".format(name, i + 1),
HighResolutionModule(
num_channels=num_channels,
num_filters=num_filters,
has_se=has_se,
name=name + '_' + str(i + 1)))
self.stage_func_list.append(stage_func)
def forward(self, input):
out = input
for idx in range(self._num_modules):
out = self.stage_func_list[idx](out)
return out
class HighResolutionModule(nn.Layer):
def __init__(self,
num_channels,
num_filters,
has_se=False,
multi_scale_output=True,
name=None):
super(HighResolutionModule, self).__init__()
self.branches_func = Branches(
block_num=4,
in_channels=num_channels,
out_channels=num_filters,
has_se=has_se,
name=name)
self.fuse_func = FuseLayers(
in_channels=num_filters,
out_channels=num_filters,
multi_scale_output=multi_scale_output,
name=name)
def forward(self, input):
out = self.branches_func(input)
out = self.fuse_func(out)
return out
class FuseLayers(nn.Layer):
def __init__(self,
in_channels,
out_channels,
multi_scale_output=True,
name=None):
super(FuseLayers, self).__init__()
self._actual_ch = len(in_channels) if multi_scale_output else 1
self._in_channels = in_channels
self.residual_func_list = []
for i in range(self._actual_ch):
for j in range(len(in_channels)):
residual_func = None
if j > i:
residual_func = self.add_sublayer(
"residual_{}_layer_{}_{}".format(name, i + 1, j + 1),
ConvBNLayer(
num_channels=in_channels[j],
num_filters=out_channels[i],
filter_size=1,
stride=1,
act=None,
name=name + '_layer_' + str(i + 1) + '_' +
str(j + 1)))
self.residual_func_list.append(residual_func)
elif j < i:
pre_num_filters = in_channels[j]
for k in range(i - j):
if k == i - j - 1:
residual_func = self.add_sublayer(
"residual_{}_layer_{}_{}_{}".format(
name, i + 1, j + 1, k + 1),
ConvBNLayer(
num_channels=pre_num_filters,
num_filters=out_channels[i],
filter_size=3,
stride=2,
act=None,
name=name + '_layer_' + str(i + 1) + '_' +
str(j + 1) + '_' + str(k + 1)))
pre_num_filters = out_channels[i]
else:
residual_func = self.add_sublayer(
"residual_{}_layer_{}_{}_{}".format(
name, i + 1, j + 1, k + 1),
ConvBNLayer(
num_channels=pre_num_filters,
num_filters=out_channels[j],
filter_size=3,
stride=2,
act="relu",
name=name + '_layer_' + str(i + 1) + '_' +
str(j + 1) + '_' + str(k + 1)))
pre_num_filters = out_channels[j]
self.residual_func_list.append(residual_func)
def forward(self, input):
outs = []
residual_func_idx = 0
for i in range(self._actual_ch):
residual = input[i]
for j in range(len(self._in_channels)):
if j > i:
y = self.residual_func_list[residual_func_idx](input[j])
residual_func_idx += 1
y = F.upsample(y, scale_factor=2**(j - i), mode="nearest")
residual = paddle.add(x=residual, y=y)
elif j < i:
y = input[j]
for k in range(i - j):
y = self.residual_func_list[residual_func_idx](y)
residual_func_idx += 1
residual = paddle.add(x=residual, y=y)
residual = F.relu(residual)
outs.append(residual)
return outs
class LastClsOut(nn.Layer):
def __init__(self,
num_channel_list,
has_se,
num_filters_list=[32, 64, 128, 256],
name=None):
super(LastClsOut, self).__init__()
self.func_list = []
for idx in range(len(num_channel_list)):
func = self.add_sublayer(
"conv_{}_conv_{}".format(name, idx + 1),
BottleneckBlock(
num_channels=num_channel_list[idx],
num_filters=num_filters_list[idx],
has_se=has_se,
downsample=True,
name=name + 'conv_' + str(idx + 1)))
self.func_list.append(func)
def forward(self, inputs):
outs = []
for idx, input in enumerate(inputs):
out = self.func_list[idx](input)
outs.append(out)
return outs
class HRNet(nn.Layer):
def __init__(self, width=18, has_se=False, class_dim=1000):
super(HRNet, self).__init__()
self.width = width
self.has_se = has_se
self.channels = {
18: [[18, 36], [18, 36, 72], [18, 36, 72, 144]],
30: [[30, 60], [30, 60, 120], [30, 60, 120, 240]],
32: [[32, 64], [32, 64, 128], [32, 64, 128, 256]],
40: [[40, 80], [40, 80, 160], [40, 80, 160, 320]],
44: [[44, 88], [44, 88, 176], [44, 88, 176, 352]],
48: [[48, 96], [48, 96, 192], [48, 96, 192, 384]],
60: [[60, 120], [60, 120, 240], [60, 120, 240, 480]],
64: [[64, 128], [64, 128, 256], [64, 128, 256, 512]]
}
self._class_dim = class_dim
channels_2, channels_3, channels_4 = self.channels[width]
num_modules_2, num_modules_3, num_modules_4 = 1, 4, 3
self.conv_layer1_1 = ConvBNLayer(
num_channels=3,
num_filters=64,
filter_size=3,
stride=2,
act='relu',
name="layer1_1")
self.conv_layer1_2 = ConvBNLayer(
num_channels=64,
num_filters=64,
filter_size=3,
stride=2,
act='relu',
name="layer1_2")
self.la1 = Layer1(num_channels=64, has_se=has_se, name="layer2")
self.tr1 = TransitionLayer(
in_channels=[256], out_channels=channels_2, name="tr1")
self.st2 = Stage(
num_channels=channels_2,
num_modules=num_modules_2,
num_filters=channels_2,
has_se=self.has_se,
name="st2")
self.tr2 = TransitionLayer(
in_channels=channels_2, out_channels=channels_3, name="tr2")
self.st3 = Stage(
num_channels=channels_3,
num_modules=num_modules_3,
num_filters=channels_3,
has_se=self.has_se,
name="st3")
self.tr3 = TransitionLayer(
in_channels=channels_3, out_channels=channels_4, name="tr3")
self.st4 = Stage(
num_channels=channels_4,
num_modules=num_modules_4,
num_filters=channels_4,
has_se=self.has_se,
name="st4")
# classification
num_filters_list = [32, 64, 128, 256]
self.last_cls = LastClsOut(
num_channel_list=channels_4,
has_se=self.has_se,
num_filters_list=num_filters_list,
name="cls_head", )
last_num_filters = [256, 512, 1024]
self.cls_head_conv_list = []
for idx in range(3):
self.cls_head_conv_list.append(
self.add_sublayer(
"cls_head_add{}".format(idx + 1),
ConvBNLayer(
num_channels=num_filters_list[idx] * 4,
num_filters=last_num_filters[idx],
filter_size=3,
stride=2,
name="cls_head_add" + str(idx + 1))))
self.conv_last = ConvBNLayer(
num_channels=1024,
num_filters=2048,
filter_size=1,
stride=1,
name="cls_head_last_conv")
self.pool2d_avg = AdaptiveAvgPool2D(1)
stdv = 1.0 / math.sqrt(2048 * 1.0)
self.out = Linear(
2048,
class_dim,
weight_attr=ParamAttr(
initializer=Uniform(-stdv, stdv), name="fc_weights"),
bias_attr=ParamAttr(name="fc_offset"))
def forward(self, input):
conv1 = self.conv_layer1_1(input)
conv2 = self.conv_layer1_2(conv1)
la1 = self.la1(conv2)
tr1 = self.tr1([la1])
st2 = self.st2(tr1)
tr2 = self.tr2(st2)
st3 = self.st3(tr2)
tr3 = self.tr3(st3)
st4 = self.st4(tr3)
last_cls = self.last_cls(st4)
y = last_cls[0]
for idx in range(3):
y = paddle.add(last_cls[idx + 1], self.cls_head_conv_list[idx](y))
y = self.conv_last(y)
y = self.pool2d_avg(y)
y = paddle.reshape(y, shape=[-1, y.shape[1]])
y = self.out(y)
return y
def _load_pretrained(pretrained, model, model_url, use_ssld=False):
if pretrained is False:
pass
elif pretrained is True:
load_dygraph_pretrain_from_url(model, model_url, use_ssld=use_ssld)
elif isinstance(pretrained, str):
load_dygraph_pretrain(model, pretrained)
else:
raise RuntimeError(
"pretrained type is not available. Please use `string` or `boolean` type."
)
def HRNet_W18_C(pretrained=False, use_ssld=False, **kwarg):
model = HRNet(width=18, **kwarg)
_load_pretrained(pretrained, model, MODEL_URLS["HRNet_W18_C"], use_ssld=use_ssld)
return model
def HRNet_W30_C(pretrained=False, use_ssld=False, **kwarg):
model = HRNet(width=30, **kwarg)
_load_pretrained(pretrained, model, MODEL_URLS["HRNet_W30_C"], use_ssld=use_ssld)
return model
def HRNet_W32_C(pretrained=False, use_ssld=False, **kwarg):
model = HRNet(width=32, **kwarg)
_load_pretrained(pretrained, model, MODEL_URLS["HRNet_W32_C"], use_ssld=use_ssld)
return model
def HRNet_W40_C(pretrained=False, use_ssld=False, **kwarg):
model = HRNet(width=40, **kwarg)
_load_pretrained(pretrained, model, MODEL_URLS["HRNet_W40_C"], use_ssld=use_ssld)
return model
def HRNet_W44_C(pretrained=False, use_ssld=False, **kwarg):
model = HRNet(width=44, **kwarg)
_load_pretrained(pretrained, model, MODEL_URLS["HRNet_W44_C"], use_ssld=use_ssld)
return model
def HRNet_W48_C(pretrained=False, use_ssld=False, **kwarg):
model = HRNet(width=48, **kwarg)
_load_pretrained(pretrained, model, MODEL_URLS["HRNet_W48_C"], use_ssld=use_ssld)
return model
def HRNet_W64_C(pretrained=False, use_ssld=False, **kwarg):
model = HRNet(width=64, **kwarg)
_load_pretrained(pretrained, model, MODEL_URLS["HRNet_W64_C"], use_ssld=use_ssld)
return model
def SE_HRNet_W64_C(pretrained=False, use_ssld=False, **kwarg):
model = HRNet(width=64, **kwarg)
_load_pretrained(pretrained, model, MODEL_URLS["SE_HRNet_W64_C"], use_ssld=use_ssld)
return model
ppcls/arch/backbone/model_zoo/inception_v3.py 已删除 100644 → 0
浏览文件 @ 3579f5a6
# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import paddle
from paddle import ParamAttr
import paddle.nn as nn
import paddle.nn.functional as F
from paddle.nn import Conv2D, BatchNorm, Linear, Dropout
from paddle.nn import AdaptiveAvgPool2D, MaxPool2D, AvgPool2D
from paddle.nn.initializer import Uniform
import math
from ppcls.utils.save_load import load_dygraph_pretrain, load_dygraph_pretrain_from_url
MODEL_URLS = {"InceptionV3": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/InceptionV3_pretrained.pdparams"}
__all__ = list(MODEL_URLS.keys())
class ConvBNLayer(nn.Layer):
def __init__(self,
num_channels,
num_filters,
filter_size,
stride=1,
padding=0,
groups=1,
act="relu",
name=None):
super(ConvBNLayer, self).__init__()
self.conv = Conv2D(
in_channels=num_channels,
out_channels=num_filters,
kernel_size=filter_size,
stride=stride,
padding=padding,
groups=groups,
weight_attr=ParamAttr(name=name+"_weights"),
bias_attr=False)
self.batch_norm = BatchNorm(
num_filters,
act=act,
param_attr=ParamAttr(name=name+"_bn_scale"),
bias_attr=ParamAttr(name=name+"_bn_offset"),
moving_mean_name=name+"_bn_mean",
moving_variance_name=name+"_bn_variance")
def forward(self, inputs):
y = self.conv(inputs)
y = self.batch_norm(y)
return y
class InceptionStem(nn.Layer):
def __init__(self):
super(InceptionStem, self).__init__()
self.conv_1a_3x3 = ConvBNLayer(num_channels=3,
num_filters=32,
filter_size=3,
stride=2,
act="relu",
name="conv_1a_3x3")
self.conv_2a_3x3 = ConvBNLayer(num_channels=32,
num_filters=32,
filter_size=3,
stride=1,
act="relu",
name="conv_2a_3x3")
self.conv_2b_3x3 = ConvBNLayer(num_channels=32,
num_filters=64,
filter_size=3,
padding=1,
act="relu",
name="conv_2b_3x3")
self.maxpool = MaxPool2D(kernel_size=3, stride=2, padding=0)
self.conv_3b_1x1 = ConvBNLayer(num_channels=64,
num_filters=80,
filter_size=1,
act="relu",
name="conv_3b_1x1")
self.conv_4a_3x3 = ConvBNLayer(num_channels=80,
num_filters=192,
filter_size=3,
act="relu",
name="conv_4a_3x3")
def forward(self, x):
y = self.conv_1a_3x3(x)
y = self.conv_2a_3x3(y)
y = self.conv_2b_3x3(y)
y = self.maxpool(y)
y = self.conv_3b_1x1(y)
y = self.conv_4a_3x3(y)
y = self.maxpool(y)
return y
class InceptionA(nn.Layer):
def __init__(self, num_channels, pool_features, name=None):
super(InceptionA, self).__init__()
self.branch1x1 = ConvBNLayer(num_channels=num_channels,
num_filters=64,
filter_size=1,
act="relu",
name="inception_a_branch1x1_"+name)
self.branch5x5_1 = ConvBNLayer(num_channels=num_channels,
num_filters=48,
filter_size=1,
act="relu",
name="inception_a_branch5x5_1_"+name)
self.branch5x5_2 = ConvBNLayer(num_channels=48,
num_filters=64,
filter_size=5,
padding=2,
act="relu",
name="inception_a_branch5x5_2_"+name)
self.branch3x3dbl_1 = ConvBNLayer(num_channels=num_channels,
num_filters=64,
filter_size=1,
act="relu",
name="inception_a_branch3x3dbl_1_"+name)
self.branch3x3dbl_2 = ConvBNLayer(num_channels=64,
num_filters=96,
filter_size=3,
padding=1,
act="relu",
name="inception_a_branch3x3dbl_2_"+name)
self.branch3x3dbl_3 = ConvBNLayer(num_channels=96,
num_filters=96,
filter_size=3,
padding=1,
act="relu",
name="inception_a_branch3x3dbl_3_"+name)
self.branch_pool = AvgPool2D(kernel_size=3, stride=1, padding=1, exclusive=False)
self.branch_pool_conv = ConvBNLayer(num_channels=num_channels,
num_filters=pool_features,
filter_size=1,
act="relu",
name="inception_a_branch_pool_"+name)
def forward(self, x):
branch1x1 = self.branch1x1(x)
branch5x5 = self.branch5x5_1(x)
branch5x5 = self.branch5x5_2(branch5x5)
branch3x3dbl = self.branch3x3dbl_1(x)
branch3x3dbl = self.branch3x3dbl_2(branch3x3dbl)
branch3x3dbl = self.branch3x3dbl_3(branch3x3dbl)
branch_pool = self.branch_pool(x)
branch_pool = self.branch_pool_conv(branch_pool)
outputs = paddle.concat([branch1x1, branch5x5, branch3x3dbl, branch_pool], axis=1)
return outputs
class InceptionB(nn.Layer):
def __init__(self, num_channels, name=None):
super(InceptionB, self).__init__()
self.branch3x3 = ConvBNLayer(num_channels=num_channels,
num_filters=384,
filter_size=3,
stride=2,
act="relu",
name="inception_b_branch3x3_"+name)
self.branch3x3dbl_1 = ConvBNLayer(num_channels=num_channels,
num_filters=64,
filter_size=1,
act="relu",
name="inception_b_branch3x3dbl_1_"+name)
self.branch3x3dbl_2 = ConvBNLayer(num_channels=64,
num_filters=96,
filter_size=3,
padding=1,
act="relu",
name="inception_b_branch3x3dbl_2_"+name)
self.branch3x3dbl_3 = ConvBNLayer(num_channels=96,
num_filters=96,
filter_size=3,
stride=2,
act="relu",
name="inception_b_branch3x3dbl_3_"+name)
self.branch_pool = MaxPool2D(kernel_size=3, stride=2)
def forward(self, x):
branch3x3 = self.branch3x3(x)
branch3x3dbl = self.branch3x3dbl_1(x)
branch3x3dbl = self.branch3x3dbl_2(branch3x3dbl)
branch3x3dbl = self.branch3x3dbl_3(branch3x3dbl)
branch_pool = self.branch_pool(x)
outputs = paddle.concat([branch3x3, branch3x3dbl, branch_pool], axis=1)
return outputs
class InceptionC(nn.Layer):
def __init__(self, num_channels, channels_7x7, name=None):
super(InceptionC, self).__init__()
self.branch1x1 = ConvBNLayer(num_channels=num_channels,
num_filters=192,
filter_size=1,
act="relu",
name="inception_c_branch1x1_"+name)
self.branch7x7_1 = ConvBNLayer(num_channels=num_channels,
num_filters=channels_7x7,
filter_size=1,
stride=1,
act="relu",
name="inception_c_branch7x7_1_"+name)
self.branch7x7_2 = ConvBNLayer(num_channels=channels_7x7,
num_filters=channels_7x7,
filter_size=(1, 7),
stride=1,
padding=(0, 3),
act="relu",
name="inception_c_branch7x7_2_"+name)
self.branch7x7_3 = ConvBNLayer(num_channels=channels_7x7,
num_filters=192,
filter_size=(7, 1),
stride=1,
padding=(3, 0),
act="relu",
name="inception_c_branch7x7_3_"+name)
self.branch7x7dbl_1 = ConvBNLayer(num_channels=num_channels,
num_filters=channels_7x7,
filter_size=1,
act="relu",
name="inception_c_branch7x7dbl_1_"+name)
self.branch7x7dbl_2 = ConvBNLayer(num_channels=channels_7x7,
num_filters=channels_7x7,
filter_size=(7, 1),
padding = (3, 0),
act="relu",
name="inception_c_branch7x7dbl_2_"+name)
self.branch7x7dbl_3 = ConvBNLayer(num_channels=channels_7x7,
num_filters=channels_7x7,
filter_size=(1, 7),
padding = (0, 3),
act="relu",
name="inception_c_branch7x7dbl_3_"+name)
self.branch7x7dbl_4 = ConvBNLayer(num_channels=channels_7x7,
num_filters=channels_7x7,
filter_size=(7, 1),
padding = (3, 0),
act="relu",
name="inception_c_branch7x7dbl_4_"+name)
self.branch7x7dbl_5 = ConvBNLayer(num_channels=channels_7x7,
num_filters=192,
filter_size=(1, 7),
padding = (0, 3),
act="relu",
name="inception_c_branch7x7dbl_5_"+name)
self.branch_pool = AvgPool2D(kernel_size=3, stride=1, padding=1, exclusive=False)
self.branch_pool_conv = ConvBNLayer(num_channels=num_channels,
num_filters=192,
filter_size=1,
act="relu",
name="inception_c_branch_pool_"+name)
def forward(self, x):
branch1x1 = self.branch1x1(x)
branch7x7 = self.branch7x7_1(x)
branch7x7 = self.branch7x7_2(branch7x7)
branch7x7 = self.branch7x7_3(branch7x7)
branch7x7dbl = self.branch7x7dbl_1(x)
branch7x7dbl = self.branch7x7dbl_2(branch7x7dbl)
branch7x7dbl = self.branch7x7dbl_3(branch7x7dbl)
branch7x7dbl = self.branch7x7dbl_4(branch7x7dbl)
branch7x7dbl = self.branch7x7dbl_5(branch7x7dbl)
branch_pool = self.branch_pool(x)
branch_pool = self.branch_pool_conv(branch_pool)
outputs = paddle.concat([branch1x1, branch7x7, branch7x7dbl, branch_pool], axis=1)
return outputs
class InceptionD(nn.Layer):
def __init__(self, num_channels, name=None):
super(InceptionD, self).__init__()
self.branch3x3_1 = ConvBNLayer(num_channels=num_channels,
num_filters=192,
filter_size=1,
act="relu",
name="inception_d_branch3x3_1_"+name)
self.branch3x3_2 = ConvBNLayer(num_channels=192,
num_filters=320,
filter_size=3,
stride=2,
act="relu",
name="inception_d_branch3x3_2_"+name)
self.branch7x7x3_1 = ConvBNLayer(num_channels=num_channels,
num_filters=192,
filter_size=1,
act="relu",
name="inception_d_branch7x7x3_1_"+name)
self.branch7x7x3_2 = ConvBNLayer(num_channels=192,
num_filters=192,
filter_size=(1, 7),
padding=(0, 3),
act="relu",
name="inception_d_branch7x7x3_2_"+name)
self.branch7x7x3_3 = ConvBNLayer(num_channels=192,
num_filters=192,
filter_size=(7, 1),
padding=(3, 0),
act="relu",
name="inception_d_branch7x7x3_3_"+name)
self.branch7x7x3_4 = ConvBNLayer(num_channels=192,
num_filters=192,
filter_size=3,
stride=2,
act="relu",
name="inception_d_branch7x7x3_4_"+name)
self.branch_pool = MaxPool2D(kernel_size=3, stride=2)
def forward(self, x):
branch3x3 = self.branch3x3_1(x)
branch3x3 = self.branch3x3_2(branch3x3)
branch7x7x3 = self.branch7x7x3_1(x)
branch7x7x3 = self.branch7x7x3_2(branch7x7x3)
branch7x7x3 = self.branch7x7x3_3(branch7x7x3)
branch7x7x3 = self.branch7x7x3_4(branch7x7x3)
branch_pool = self.branch_pool(x)
outputs = paddle.concat([branch3x3, branch7x7x3, branch_pool], axis=1)
return outputs
class InceptionE(nn.Layer):
def __init__(self, num_channels, name=None):
super(InceptionE, self).__init__()
self.branch1x1 = ConvBNLayer(num_channels=num_channels,
num_filters=320,
filter_size=1,
act="relu",
name="inception_e_branch1x1_"+name)
self.branch3x3_1 = ConvBNLayer(num_channels=num_channels,
num_filters=384,
filter_size=1,
act="relu",
name="inception_e_branch3x3_1_"+name)
self.branch3x3_2a = ConvBNLayer(num_channels=384,
num_filters=384,
filter_size=(1, 3),
padding=(0, 1),
act="relu",
name="inception_e_branch3x3_2a_"+name)
self.branch3x3_2b = ConvBNLayer(num_channels=384,
num_filters=384,
filter_size=(3, 1),
padding=(1, 0),
act="relu",
name="inception_e_branch3x3_2b_"+name)
self.branch3x3dbl_1 = ConvBNLayer(num_channels=num_channels,
num_filters=448,
filter_size=1,
act="relu",
name="inception_e_branch3x3dbl_1_"+name)
self.branch3x3dbl_2 = ConvBNLayer(num_channels=448,
num_filters=384,
filter_size=3,
padding=1,
act="relu",
name="inception_e_branch3x3dbl_2_"+name)
self.branch3x3dbl_3a = ConvBNLayer(num_channels=384,
num_filters=384,
filter_size=(1, 3),
padding=(0, 1),
act="relu",
name="inception_e_branch3x3dbl_3a_"+name)
self.branch3x3dbl_3b = ConvBNLayer(num_channels=384,
num_filters=384,
filter_size=(3, 1),
padding=(1, 0),
act="relu",
name="inception_e_branch3x3dbl_3b_"+name)
self.branch_pool = AvgPool2D(kernel_size=3, stride=1, padding=1, exclusive=False)
self.branch_pool_conv = ConvBNLayer(num_channels=num_channels,
num_filters=192,
filter_size=1,
act="relu",
name="inception_e_branch_pool_"+name)
def forward(self, x):
branch1x1 = self.branch1x1(x)
branch3x3 = self.branch3x3_1(x)
branch3x3 = [
self.branch3x3_2a(branch3x3),
self.branch3x3_2b(branch3x3),
]
branch3x3 = paddle.concat(branch3x3, axis=1)
branch3x3dbl = self.branch3x3dbl_1(x)
branch3x3dbl = self.branch3x3dbl_2(branch3x3dbl)
branch3x3dbl = [
self.branch3x3dbl_3a(branch3x3dbl),
self.branch3x3dbl_3b(branch3x3dbl),
]
branch3x3dbl = paddle.concat(branch3x3dbl, axis=1)
branch_pool = self.branch_pool(x)
branch_pool = self.branch_pool_conv(branch_pool)
outputs = paddle.concat([branch1x1, branch3x3, branch3x3dbl, branch_pool], axis=1)
return outputs
class Inception_V3(nn.Layer):
def __init__(self, class_dim=1000):
super(Inception_V3, self).__init__()
self.inception_a_list = [[192, 256, 288], [32, 64, 64]]
self.inception_c_list = [[768, 768, 768, 768], [128, 160, 160, 192]]
self.inception_stem = InceptionStem()
self.inception_block_list = []
for i in range(len(self.inception_a_list[0])):
inception_a = self.add_sublayer("inception_a_"+str(i+1),
InceptionA(self.inception_a_list[0][i],
self.inception_a_list[1][i],
name=str(i+1)))
self.inception_block_list.append(inception_a)
inception_b = self.add_sublayer("nception_b_1",
InceptionB(288, name="1"))
self.inception_block_list.append(inception_b)
for i in range(len(self.inception_c_list[0])):
inception_c = self.add_sublayer("inception_c_"+str(i+1),
InceptionC(self.inception_c_list[0][i],
self.inception_c_list[1][i],
name=str(i+1)))
self.inception_block_list.append(inception_c)
inception_d = self.add_sublayer("inception_d_1",
InceptionD(768, name="1"))
self.inception_block_list.append(inception_d)
inception_e = self.add_sublayer("inception_e_1",
InceptionE(1280, name="1"))
self.inception_block_list.append(inception_e)
inception_e = self.add_sublayer("inception_e_2",
InceptionE(2048, name="2"))
self.inception_block_list.append(inception_e)
self.gap = AdaptiveAvgPool2D(1)
self.drop = Dropout(p=0.2, mode="downscale_in_infer")
stdv = 1.0 / math.sqrt(2048 * 1.0)
self.out = Linear(
2048,
class_dim,
weight_attr=ParamAttr(
initializer=Uniform(-stdv, stdv), name="fc_weights"),
bias_attr=ParamAttr(name="fc_offset"))
def forward(self, x):
y = self.inception_stem(x)
for inception_block in self.inception_block_list:
y = inception_block(y)
y = self.gap(y)
y = paddle.reshape(y, shape=[-1, 2048])
y = self.drop(y)
y = self.out(y)
return y
def _load_pretrained(pretrained, model, model_url, use_ssld=False):
if pretrained is False:
pass
elif pretrained is True:
load_dygraph_pretrain_from_url(model, model_url, use_ssld=use_ssld)
elif isinstance(pretrained, str):
load_dygraph_pretrain(model, pretrained)
else:
raise RuntimeError(
"pretrained type is not available. Please use `string` or `boolean` type."
)
def InceptionV3(pretrained=False, use_ssld=False, **kwargs):
model = Inception_V3(**kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["InceptionV3"], use_ssld=use_ssld)
return model
ppcls/arch/backbone/model_zoo/mobilenet_v1.py 已删除 100644 → 0
浏览文件 @ 3579f5a6
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import numpy as np
import paddle
from paddle import ParamAttr
import paddle.nn as nn
import paddle.nn.functional as F
from paddle.nn import Conv2D, BatchNorm, Linear, Dropout
from paddle.nn import AdaptiveAvgPool2D, MaxPool2D, AvgPool2D
from paddle.nn.initializer import KaimingNormal
import math
from ppcls.utils.save_load import load_dygraph_pretrain, load_dygraph_pretrain_from_url
MODEL_URLS = {"MobileNetV1_x0_25": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_x0_25_pretrained.pdparams",
"MobileNetV1_x0_5": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_x0_5_pretrained.pdparams",
"MobileNetV1_x0_75": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_x0_75_pretrained.pdparams",
"MobileNetV1": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV1_pretrained.pdparams"}
__all__ = list(MODEL_URLS.keys())
class ConvBNLayer(nn.Layer):
def __init__(self,
num_channels,
filter_size,
num_filters,
stride,
padding,
channels=None,
num_groups=1,
act='relu',
name=None):
super(ConvBNLayer, self).__init__()
self._conv = Conv2D(
in_channels=num_channels,
out_channels=num_filters,
kernel_size=filter_size,
stride=stride,
padding=padding,
groups=num_groups,
weight_attr=ParamAttr(
initializer=KaimingNormal(), name=name + "_weights"),
bias_attr=False)
self._batch_norm = BatchNorm(
num_filters,
act=act,
param_attr=ParamAttr(name + "_bn_scale"),
bias_attr=ParamAttr(name + "_bn_offset"),
moving_mean_name=name + "_bn_mean",
moving_variance_name=name + "_bn_variance")
def forward(self, inputs):
y = self._conv(inputs)
y = self._batch_norm(y)
return y
class DepthwiseSeparable(nn.Layer):
def __init__(self,
num_channels,
num_filters1,
num_filters2,
num_groups,
stride,
scale,
name=None):
super(DepthwiseSeparable, self).__init__()
self._depthwise_conv = ConvBNLayer(
num_channels=num_channels,
num_filters=int(num_filters1 * scale),
filter_size=3,
stride=stride,
padding=1,
num_groups=int(num_groups * scale),
name=name + "_dw")
self._pointwise_conv = ConvBNLayer(
num_channels=int(num_filters1 * scale),
filter_size=1,
num_filters=int(num_filters2 * scale),
stride=1,
padding=0,
name=name + "_sep")
def forward(self, inputs):
y = self._depthwise_conv(inputs)
y = self._pointwise_conv(y)
return y
class MobileNet(nn.Layer):
def __init__(self, scale=1.0, class_dim=1000):
super(MobileNet, self).__init__()
self.scale = scale
self.block_list = []
self.conv1 = ConvBNLayer(
num_channels=3,
filter_size=3,
channels=3,
num_filters=int(32 * scale),
stride=2,
padding=1,
name="conv1")
conv2_1 = self.add_sublayer(
"conv2_1",
sublayer=DepthwiseSeparable(
num_channels=int(32 * scale),
num_filters1=32,
num_filters2=64,
num_groups=32,
stride=1,
scale=scale,
name="conv2_1"))
self.block_list.append(conv2_1)
conv2_2 = self.add_sublayer(
"conv2_2",
sublayer=DepthwiseSeparable(
num_channels=int(64 * scale),
num_filters1=64,
num_filters2=128,
num_groups=64,
stride=2,
scale=scale,
name="conv2_2"))
self.block_list.append(conv2_2)
conv3_1 = self.add_sublayer(
"conv3_1",
sublayer=DepthwiseSeparable(
num_channels=int(128 * scale),
num_filters1=128,
num_filters2=128,
num_groups=128,
stride=1,
scale=scale,
name="conv3_1"))
self.block_list.append(conv3_1)
conv3_2 = self.add_sublayer(
"conv3_2",
sublayer=DepthwiseSeparable(
num_channels=int(128 * scale),
num_filters1=128,
num_filters2=256,
num_groups=128,
stride=2,
scale=scale,
name="conv3_2"))
self.block_list.append(conv3_2)
conv4_1 = self.add_sublayer(
"conv4_1",
sublayer=DepthwiseSeparable(
num_channels=int(256 * scale),
num_filters1=256,
num_filters2=256,
num_groups=256,
stride=1,
scale=scale,
name="conv4_1"))
self.block_list.append(conv4_1)
conv4_2 = self.add_sublayer(
"conv4_2",
sublayer=DepthwiseSeparable(
num_channels=int(256 * scale),
num_filters1=256,
num_filters2=512,
num_groups=256,
stride=2,
scale=scale,
name="conv4_2"))
self.block_list.append(conv4_2)
for i in range(5):
conv5 = self.add_sublayer(
"conv5_" + str(i + 1),
sublayer=DepthwiseSeparable(
num_channels=int(512 * scale),
num_filters1=512,
num_filters2=512,
num_groups=512,
stride=1,
scale=scale,
name="conv5_" + str(i + 1)))
self.block_list.append(conv5)
conv5_6 = self.add_sublayer(
"conv5_6",
sublayer=DepthwiseSeparable(
num_channels=int(512 * scale),
num_filters1=512,
num_filters2=1024,
num_groups=512,
stride=2,
scale=scale,
name="conv5_6"))
self.block_list.append(conv5_6)
conv6 = self.add_sublayer(
"conv6",
sublayer=DepthwiseSeparable(
num_channels=int(1024 * scale),
num_filters1=1024,
num_filters2=1024,
num_groups=1024,
stride=1,
scale=scale,
name="conv6"))
self.block_list.append(conv6)
self.pool2d_avg = AdaptiveAvgPool2D(1)
self.out = Linear(
int(1024 * scale),
class_dim,
weight_attr=ParamAttr(
initializer=KaimingNormal(), name="fc7_weights"),
bias_attr=ParamAttr(name="fc7_offset"))
def forward(self, inputs):
y = self.conv1(inputs)
for block in self.block_list:
y = block(y)
y = self.pool2d_avg(y)
y = paddle.flatten(y, start_axis=1, stop_axis=-1)
y = self.out(y)
return y
def _load_pretrained(pretrained, model, model_url, use_ssld=False):
if pretrained is False:
pass
elif pretrained is True:
load_dygraph_pretrain_from_url(model, model_url, use_ssld=use_ssld)
elif isinstance(pretrained, str):
load_dygraph_pretrain(model, pretrained)
else:
raise RuntimeError(
"pretrained type is not available. Please use `string` or `boolean` type."
)
def MobileNetV1_x0_25(pretrained=False, use_ssld=False, **kwargs):
model = MobileNet(scale=0.25, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV1_x0_25"], use_ssld=use_ssld)
return model
def MobileNetV1_x0_5(pretrained=False, use_ssld=False, **kwargs):
model = MobileNet(scale=0.5, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV1_x0_5"], use_ssld=use_ssld)
return model
def MobileNetV1_x0_75(pretrained=False, use_ssld=False, **kwargs):
model = MobileNet(scale=0.75, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV1_x0_75"], use_ssld=use_ssld)
return model
def MobileNetV1(pretrained=False, use_ssld=False, **kwargs):
model = MobileNet(scale=1.0, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV1"], use_ssld=use_ssld)
return model
\ No newline at end of file
ppcls/arch/backbone/model_zoo/mobilenet_v3.py 已删除 100644 → 0
浏览文件 @ 3579f5a6
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import numpy as np
import paddle
from paddle import ParamAttr
import paddle.nn as nn
import paddle.nn.functional as F
from paddle.nn.functional import hardswish, hardsigmoid
from paddle.nn import Conv2D, BatchNorm, Linear, Dropout
from paddle.nn import AdaptiveAvgPool2D, MaxPool2D, AvgPool2D
from paddle.regularizer import L2Decay
import math
from ppcls.utils.save_load import load_dygraph_pretrain, load_dygraph_pretrain_from_url
MODEL_URLS = {"MobileNetV3_small_x0_35": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x0_35_pretrained.pdparams",
"MobileNetV3_small_x0_5": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x0_5_pretrained.pdparams",
"MobileNetV3_small_x0_75": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x0_75_pretrained.pdparams",
"MobileNetV3_small_x1_0": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x1_0_pretrained.pdparams",
"MobileNetV3_small_x1_25": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_small_x1_25_pretrained.pdparams",
"MobileNetV3_large_x0_35": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x0_35_pretrained.pdparams",
"MobileNetV3_large_x0_5": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x0_5_pretrained.pdparams",
"MobileNetV3_large_x0_75": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x0_75_pretrained.pdparams",
"MobileNetV3_large_x1_0": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x1_0_pretrained.pdparams",
"MobileNetV3_large_x1_25": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/MobileNetV3_large_x1_25_pretrained.pdparams"}
__all__ = list(MODEL_URLS.keys())
def make_divisible(v, divisor=8, min_value=None):
if min_value is None:
min_value = divisor
new_v = max(min_value, int(v + divisor / 2) // divisor * divisor)
if new_v < 0.9 * v:
new_v += divisor
return new_v
class MobileNetV3(nn.Layer):
def __init__(self,
scale=1.0,
model_name="small",
dropout_prob=0.2,
class_dim=1000):
super(MobileNetV3, self).__init__()
inplanes = 16
if model_name == "large":
self.cfg = [
# k, exp, c, se, nl, s,
[3, 16, 16, False, "relu", 1],
[3, 64, 24, False, "relu", 2],
[3, 72, 24, False, "relu", 1],
[5, 72, 40, True, "relu", 2],
[5, 120, 40, True, "relu", 1],
[5, 120, 40, True, "relu", 1],
[3, 240, 80, False, "hardswish", 2],
[3, 200, 80, False, "hardswish", 1],
[3, 184, 80, False, "hardswish", 1],
[3, 184, 80, False, "hardswish", 1],
[3, 480, 112, True, "hardswish", 1],
[3, 672, 112, True, "hardswish", 1],
[5, 672, 160, True, "hardswish", 2],
[5, 960, 160, True, "hardswish", 1],
[5, 960, 160, True, "hardswish", 1],
]
self.cls_ch_squeeze = 960
self.cls_ch_expand = 1280
elif model_name == "small":
self.cfg = [
# k, exp, c, se, nl, s,
[3, 16, 16, True, "relu", 2],
[3, 72, 24, False, "relu", 2],
[3, 88, 24, False, "relu", 1],
[5, 96, 40, True, "hardswish", 2],
[5, 240, 40, True, "hardswish", 1],
[5, 240, 40, True, "hardswish", 1],
[5, 120, 48, True, "hardswish", 1],
[5, 144, 48, True, "hardswish", 1],
[5, 288, 96, True, "hardswish", 2],
[5, 576, 96, True, "hardswish", 1],
[5, 576, 96, True, "hardswish", 1],
]
self.cls_ch_squeeze = 576
self.cls_ch_expand = 1280
else:
raise NotImplementedError(
"mode[{}_model] is not implemented!".format(model_name))
self.conv1 = ConvBNLayer(
in_c=3,
out_c=make_divisible(inplanes * scale),
filter_size=3,
stride=2,
padding=1,
num_groups=1,
if_act=True,
act="hardswish",
name="conv1")
self.block_list = []
i = 0
inplanes = make_divisible(inplanes * scale)
for (k, exp, c, se, nl, s) in self.cfg:
block = self.add_sublayer(
"conv" + str(i + 2),
ResidualUnit(
in_c=inplanes,
mid_c=make_divisible(scale * exp),
out_c=make_divisible(scale * c),
filter_size=k,
stride=s,
use_se=se,
act=nl,
name="conv" + str(i + 2)))
self.block_list.append(block)
inplanes = make_divisible(scale * c)
i += 1
self.last_second_conv = ConvBNLayer(
in_c=inplanes,
out_c=make_divisible(scale * self.cls_ch_squeeze),
filter_size=1,
stride=1,
padding=0,
num_groups=1,
if_act=True,
act="hardswish",
name="conv_last")
self.pool = AdaptiveAvgPool2D(1)
self.last_conv = Conv2D(
in_channels=make_divisible(scale * self.cls_ch_squeeze),
out_channels=self.cls_ch_expand,
kernel_size=1,
stride=1,
padding=0,
weight_attr=ParamAttr(name="last_1x1_conv_weights"),
bias_attr=False)
self.dropout = Dropout(p=dropout_prob, mode="downscale_in_infer")
self.out = Linear(
self.cls_ch_expand,
class_dim,
weight_attr=ParamAttr("fc_weights"),
bias_attr=ParamAttr(name="fc_offset"))
def forward(self, inputs):
x = self.conv1(inputs)
for block in self.block_list:
x = block(x)
x = self.last_second_conv(x)
x = self.pool(x)
x = self.last_conv(x)
x = hardswish(x)
x = self.dropout(x)
x = paddle.flatten(x, start_axis=1, stop_axis=-1)
x = self.out(x)
return x
class ConvBNLayer(nn.Layer):
def __init__(self,
in_c,
out_c,
filter_size,
stride,
padding,
num_groups=1,
if_act=True,
act=None,
use_cudnn=True,
name=""):
super(ConvBNLayer, self).__init__()
self.if_act = if_act
self.act = act
self.conv = Conv2D(
in_channels=in_c,
out_channels=out_c,
kernel_size=filter_size,
stride=stride,
padding=padding,
groups=num_groups,
weight_attr=ParamAttr(name=name + "_weights"),
bias_attr=False)
self.bn = BatchNorm(
num_channels=out_c,
act=None,
param_attr=ParamAttr(
name=name + "_bn_scale", regularizer=L2Decay(0.0)),
bias_attr=ParamAttr(
name=name + "_bn_offset", regularizer=L2Decay(0.0)),
moving_mean_name=name + "_bn_mean",
moving_variance_name=name + "_bn_variance")
def forward(self, x):
x = self.conv(x)
x = self.bn(x)
if self.if_act:
if self.act == "relu":
x = F.relu(x)
elif self.act == "hardswish":
x = hardswish(x)
else:
print("The activation function is selected incorrectly.")
exit()
return x
class ResidualUnit(nn.Layer):
def __init__(self,
in_c,
mid_c,
out_c,
filter_size,
stride,
use_se,
act=None,
name=''):
super(ResidualUnit, self).__init__()
self.if_shortcut = stride == 1 and in_c == out_c
self.if_se = use_se
self.expand_conv = ConvBNLayer(
in_c=in_c,
out_c=mid_c,
filter_size=1,
stride=1,
padding=0,
if_act=True,
act=act,
name=name + "_expand")
self.bottleneck_conv = ConvBNLayer(
in_c=mid_c,
out_c=mid_c,
filter_size=filter_size,
stride=stride,
padding=int((filter_size - 1) // 2),
num_groups=mid_c,
if_act=True,
act=act,
name=name + "_depthwise")
if self.if_se:
self.mid_se = SEModule(mid_c, name=name + "_se")
self.linear_conv = ConvBNLayer(
in_c=mid_c,
out_c=out_c,
filter_size=1,
stride=1,
padding=0,
if_act=False,
act=None,
name=name + "_linear")
def forward(self, inputs):
x = self.expand_conv(inputs)
x = self.bottleneck_conv(x)
if self.if_se:
x = self.mid_se(x)
x = self.linear_conv(x)
if self.if_shortcut:
x = paddle.add(inputs, x)
return x
class SEModule(nn.Layer):
def __init__(self, channel, reduction=4, name=""):
super(SEModule, self).__init__()
self.avg_pool = AdaptiveAvgPool2D(1)
self.conv1 = Conv2D(
in_channels=channel,
out_channels=channel // reduction,
kernel_size=1,
stride=1,
padding=0,
weight_attr=ParamAttr(name=name + "_1_weights"),
bias_attr=ParamAttr(name=name + "_1_offset"))
self.conv2 = Conv2D(
in_channels=channel // reduction,
out_channels=channel,
kernel_size=1,
stride=1,
padding=0,
weight_attr=ParamAttr(name + "_2_weights"),
bias_attr=ParamAttr(name=name + "_2_offset"))
def forward(self, inputs):
outputs = self.avg_pool(inputs)
outputs = self.conv1(outputs)
outputs = F.relu(outputs)
outputs = self.conv2(outputs)
outputs = hardsigmoid(outputs, slope=0.2, offset=0.5)
return paddle.multiply(x=inputs, y=outputs)
def _load_pretrained(pretrained, model, model_url, use_ssld=False):
if pretrained is False:
pass
elif pretrained is True:
load_dygraph_pretrain_from_url(model, model_url, use_ssld=use_ssld)
elif isinstance(pretrained, str):
load_dygraph_pretrain(model, pretrained)
else:
raise RuntimeError(
"pretrained type is not available. Please use `string` or `boolean` type."
)
def MobileNetV3_small_x0_35(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="small", scale=0.35, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_small_x0_35"], use_ssld=use_ssld)
return model
def MobileNetV3_small_x0_5(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="small", scale=0.5, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_small_x0_5"], use_ssld=use_ssld)
return model
def MobileNetV3_small_x0_75(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="small", scale=0.75, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_small_x0_75"], use_ssld=use_ssld)
return model
def MobileNetV3_small_x1_0(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="small", scale=1.0, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_small_x1_0"], use_ssld=use_ssld)
return model
def MobileNetV3_small_x1_25(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="small", scale=1.25, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_small_x1_25"], use_ssld=use_ssld)
return model
def MobileNetV3_large_x0_35(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="large", scale=0.35, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_large_x0_35"], use_ssld=use_ssld)
return model
def MobileNetV3_large_x0_5(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="large", scale=0.5, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_large_x0_5"], use_ssld=use_ssld)
return model
def MobileNetV3_large_x0_75(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="large", scale=0.75, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_large_x0_75"], use_ssld=use_ssld)
return model
def MobileNetV3_large_x1_0(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="large", scale=1.0, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_large_x1_0"], use_ssld=use_ssld)
return model
def MobileNetV3_large_x1_25(pretrained=False, use_ssld=False, **kwargs):
model = MobileNetV3(model_name="large", scale=1.25, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["MobileNetV3_large_x1_25"], use_ssld=use_ssld)
return model
ppcls/arch/backbone/model_zoo/resnet.py 已删除 100644 → 0
浏览文件 @ 3579f5a6
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import numpy as np
import paddle
from paddle import ParamAttr
import paddle.nn as nn
import paddle.nn.functional as F
from paddle.nn import Conv2D, BatchNorm, Linear, Dropout
from paddle.nn import AdaptiveAvgPool2D, MaxPool2D, AvgPool2D
from paddle.nn.initializer import Uniform
import math
from ppcls.utils.save_load import load_dygraph_pretrain, load_dygraph_pretrain_from_url
MODEL_URLS = {"ResNet18": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet18_pretrained.pdparams",
"ResNet34": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet34_pretrained.pdparams",
"ResNet50": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_pretrained.pdparams",
"ResNet101": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet101_pretrained.pdparams",
"ResNet152": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet152_pretrained.pdparams",
}
__all__ = list(MODEL_URLS.keys())
class ConvBNLayer(nn.Layer):
def __init__(self,
num_channels,
num_filters,
filter_size,
stride=1,
groups=1,
act=None,
name=None,
data_format="NCHW"):
super(ConvBNLayer, self).__init__()
self._conv = Conv2D(
in_channels=num_channels,
out_channels=num_filters,
kernel_size=filter_size,
stride=stride,
padding=(filter_size - 1) // 2,
groups=groups,
weight_attr=ParamAttr(name=name + "_weights"),
bias_attr=False,
data_format=data_format)
if name == "conv1":
bn_name = "bn_" + name
else:
bn_name = "bn" + name[3:]
self._batch_norm = BatchNorm(
num_filters,
act=act,
param_attr=ParamAttr(name=bn_name + "_scale"),
bias_attr=ParamAttr(bn_name + "_offset"),
moving_mean_name=bn_name + "_mean",
moving_variance_name=bn_name + "_variance",
data_layout=data_format)
def forward(self, inputs):
y = self._conv(inputs)
y = self._batch_norm(y)
return y
class BottleneckBlock(nn.Layer):
def __init__(self,
num_channels,
num_filters,
stride,
shortcut=True,
name=None,
data_format="NCHW"):
super(BottleneckBlock, self).__init__()
self.conv0 = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters,
filter_size=1,
act="relu",
name=name + "_branch2a",
data_format=data_format)
self.conv1 = ConvBNLayer(
num_channels=num_filters,
num_filters=num_filters,
filter_size=3,
stride=stride,
act="relu",
name=name + "_branch2b",
data_format=data_format)
self.conv2 = ConvBNLayer(
num_channels=num_filters,
num_filters=num_filters * 4,
filter_size=1,
act=None,
name=name + "_branch2c",
data_format=data_format)
if not shortcut:
self.short = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters * 4,
filter_size=1,
stride=stride,
name=name + "_branch1",
data_format=data_format)
self.shortcut = shortcut
self._num_channels_out = num_filters * 4
def forward(self, inputs):
y = self.conv0(inputs)
conv1 = self.conv1(y)
conv2 = self.conv2(conv1)
if self.shortcut:
short = inputs
else:
short = self.short(inputs)
y = paddle.add(x=short, y=conv2)
y = F.relu(y)
return y
class BasicBlock(nn.Layer):
def __init__(self,
num_channels,
num_filters,
stride,
shortcut=True,
name=None,
data_format="NCHW"):
super(BasicBlock, self).__init__()
self.stride = stride
self.conv0 = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters,
filter_size=3,
stride=stride,
act="relu",
name=name + "_branch2a",
data_format=data_format)
self.conv1 = ConvBNLayer(
num_channels=num_filters,
num_filters=num_filters,
filter_size=3,
act=None,
name=name + "_branch2b",
data_format=data_format)
if not shortcut:
self.short = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters,
filter_size=1,
stride=stride,
name=name + "_branch1",
data_format=data_format)
self.shortcut = shortcut
def forward(self, inputs):
y = self.conv0(inputs)
conv1 = self.conv1(y)
if self.shortcut:
short = inputs
else:
short = self.short(inputs)
y = paddle.add(x=short, y=conv1)
y = F.relu(y)
return y
class ResNet(nn.Layer):
def __init__(self, layers=50, class_dim=1000, input_image_channel=3, data_format="NCHW"):
super(ResNet, self).__init__()
self.layers = layers
self.data_format = data_format
self.input_image_channel = input_image_channel
supported_layers = [18, 34, 50, 101, 152]
assert layers in supported_layers, \
"supported layers are {} but input layer is {}".format(
supported_layers, layers)
if layers == 18:
depth = [2, 2, 2, 2]
elif layers == 34 or layers == 50:
depth = [3, 4, 6, 3]
elif layers == 101:
depth = [3, 4, 23, 3]
elif layers == 152:
depth = [3, 8, 36, 3]
num_channels = [64, 256, 512,
1024] if layers >= 50 else [64, 64, 128, 256]
num_filters = [64, 128, 256, 512]
self.conv = ConvBNLayer(
num_channels=self.input_image_channel,
num_filters=64,
filter_size=7,
stride=2,
act="relu",
name="conv1",
data_format=self.data_format)
self.pool2d_max = MaxPool2D(
kernel_size=3,
stride=2,
padding=1,
data_format=self.data_format)
self.block_list = []
if layers >= 50:
for block in range(len(depth)):
shortcut = False
for i in range(depth[block]):
if layers in [101, 152] and block == 2:
if i == 0:
conv_name = "res" + str(block + 2) + "a"
else:
conv_name = "res" + str(block + 2) + "b" + str(i)
else:
conv_name = "res" + str(block + 2) + chr(97 + i)
bottleneck_block = self.add_sublayer(
conv_name,
BottleneckBlock(
num_channels=num_channels[block]
if i == 0 else num_filters[block] * 4,
num_filters=num_filters[block],
stride=2 if i == 0 and block != 0 else 1,
shortcut=shortcut,
name=conv_name,
data_format=self.data_format))
self.block_list.append(bottleneck_block)
shortcut = True
else:
for block in range(len(depth)):
shortcut = False
for i in range(depth[block]):
conv_name = "res" + str(block + 2) + chr(97 + i)
basic_block = self.add_sublayer(
conv_name,
BasicBlock(
num_channels=num_channels[block]
if i == 0 else num_filters[block],
num_filters=num_filters[block],
stride=2 if i == 0 and block != 0 else 1,
shortcut=shortcut,
name=conv_name,
data_format=self.data_format))
self.block_list.append(basic_block)
shortcut = True
self.pool2d_avg = AdaptiveAvgPool2D(1, data_format=self.data_format)
self.pool2d_avg_channels = num_channels[-1] * 2
stdv = 1.0 / math.sqrt(self.pool2d_avg_channels * 1.0)
self.out = Linear(
self.pool2d_avg_channels,
class_dim,
weight_attr=ParamAttr(
initializer=Uniform(-stdv, stdv), name="fc_0.w_0"),
bias_attr=ParamAttr(name="fc_0.b_0"))
def forward(self, inputs):
with paddle.static.amp.fp16_guard():
if self.data_format == "NHWC":
inputs = paddle.tensor.transpose(inputs, [0, 2, 3, 1])
inputs.stop_gradient = True
y = self.conv(inputs)
y = self.pool2d_max(y)
for block in self.block_list:
y = block(y)
y = self.pool2d_avg(y)
y = paddle.reshape(y, shape=[-1, self.pool2d_avg_channels])
y = self.out(y)
return y
def _load_pretrained(pretrained, model, model_url, use_ssld=False):
if pretrained is False:
pass
elif pretrained is True:
load_dygraph_pretrain_from_url(model, model_url, use_ssld=use_ssld)
elif isinstance(pretrained, str):
load_dygraph_pretrain(model, pretrained)
else:
raise RuntimeError(
"pretrained type is not available. Please use `string` or `boolean` type."
)
def ResNet18(pretrained=False, use_ssld=False, **kwargs):
model = ResNet(layers=18, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet18"], use_ssld=use_ssld)
return model
def ResNet34(pretrained=False, use_ssld=False, **kwargs):
model = ResNet(layers=34, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet34"], use_ssld=use_ssld)
return model
def ResNet50(pretrained=False, use_ssld=False, **kwargs):
model = ResNet(layers=50, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet50"], use_ssld=use_ssld)
return model
def ResNet101(pretrained=False, use_ssld=False, **kwargs):
model = ResNet(layers=101, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet101"], use_ssld=use_ssld)
return model
def ResNet152(pretrained=False, use_ssld=False, **kwargs):
model = ResNet(layers=152, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet152"], use_ssld=use_ssld)
return model
ppcls/arch/backbone/model_zoo/resnet_vd.py 已删除 100644 → 0
浏览文件 @ 3579f5a6
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import numpy as np
import paddle
from paddle import ParamAttr
import paddle.nn as nn
import paddle.nn.functional as F
from paddle.nn import Conv2D, BatchNorm, Linear, Dropout
from paddle.nn import AdaptiveAvgPool2D, MaxPool2D, AvgPool2D
from paddle.nn.initializer import Uniform
import math
from ppcls.utils.save_load import load_dygraph_pretrain, load_dygraph_pretrain_from_url
MODEL_URLS = {
"ResNet18_vd": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet18_vd_pretrained.pdparams",
"ResNet34_vd": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet34_vd_pretrained.pdparams",
"ResNet50_vd": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_pretrained.pdparams",
"ResNet101_vd": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet101_vd_pretrained.pdparams",
"ResNet152_vd": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet152_vd_pretrained.pdparams",
"ResNet200_vd": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet200_vd_pretrained.pdparams",
}
__all__ = list(MODEL_URLS.keys())
class ConvBNLayer(nn.Layer):
def __init__(self,
num_channels,
num_filters,
filter_size,
stride=1,
groups=1,
is_vd_mode=False,
act=None,
lr_mult=1.0,
name=None):
super(ConvBNLayer, self).__init__()
self.is_vd_mode = is_vd_mode
self._pool2d_avg = AvgPool2D(
kernel_size=2, stride=2, padding=0, ceil_mode=True)
self._conv = Conv2D(
in_channels=num_channels,
out_channels=num_filters,
kernel_size=filter_size,
stride=stride,
padding=(filter_size - 1) // 2,
groups=groups,
weight_attr=ParamAttr(
name=name + "_weights", learning_rate=lr_mult),
bias_attr=False)
if name == "conv1":
bn_name = "bn_" + name
else:
bn_name = "bn" + name[3:]
self._batch_norm = BatchNorm(
num_filters,
act=act,
param_attr=ParamAttr(
name=bn_name + '_scale', learning_rate=lr_mult),
bias_attr=ParamAttr(
bn_name + '_offset', learning_rate=lr_mult),
moving_mean_name=bn_name + '_mean',
moving_variance_name=bn_name + '_variance')
def forward(self, inputs):
if self.is_vd_mode:
inputs = self._pool2d_avg(inputs)
y = self._conv(inputs)
y = self._batch_norm(y)
return y
class BottleneckBlock(nn.Layer):
def __init__(self,
num_channels,
num_filters,
stride,
shortcut=True,
if_first=False,
lr_mult=1.0,
name=None):
super(BottleneckBlock, self).__init__()
self.conv0 = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters,
filter_size=1,
act='relu',
lr_mult=lr_mult,
name=name + "_branch2a")
self.conv1 = ConvBNLayer(
num_channels=num_filters,
num_filters=num_filters,
filter_size=3,
stride=stride,
act='relu',
lr_mult=lr_mult,
name=name + "_branch2b")
self.conv2 = ConvBNLayer(
num_channels=num_filters,
num_filters=num_filters * 4,
filter_size=1,
act=None,
lr_mult=lr_mult,
name=name + "_branch2c")
if not shortcut:
self.short = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters * 4,
filter_size=1,
stride=1,
is_vd_mode=False if if_first else True,
lr_mult=lr_mult,
name=name + "_branch1")
self.shortcut = shortcut
def forward(self, inputs):
y = self.conv0(inputs)
conv1 = self.conv1(y)
conv2 = self.conv2(conv1)
if self.shortcut:
short = inputs
else:
short = self.short(inputs)
y = paddle.add(x=short, y=conv2)
y = F.relu(y)
return y
class BasicBlock(nn.Layer):
def __init__(self,
num_channels,
num_filters,
stride,
shortcut=True,
if_first=False,
lr_mult=1.0,
name=None):
super(BasicBlock, self).__init__()
self.stride = stride
self.conv0 = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters,
filter_size=3,
stride=stride,
act='relu',
lr_mult=lr_mult,
name=name + "_branch2a")
self.conv1 = ConvBNLayer(
num_channels=num_filters,
num_filters=num_filters,
filter_size=3,
act=None,
lr_mult=lr_mult,
name=name + "_branch2b")
if not shortcut:
self.short = ConvBNLayer(
num_channels=num_channels,
num_filters=num_filters,
filter_size=1,
stride=1,
is_vd_mode=False if if_first else True,
lr_mult=lr_mult,
name=name + "_branch1")
self.shortcut = shortcut
def forward(self, inputs):
y = self.conv0(inputs)
conv1 = self.conv1(y)
if self.shortcut:
short = inputs
else:
short = self.short(inputs)
y = paddle.add(x=short, y=conv1)
y = F.relu(y)
return y
class ResNet_vd(nn.Layer):
def __init__(self,
layers=50,
class_dim=1000,
lr_mult_list=[1.0, 1.0, 1.0, 1.0, 1.0]):
super(ResNet_vd, self).__init__()
self.layers = layers
supported_layers = [18, 34, 50, 101, 152, 200]
assert layers in supported_layers, \
"supported layers are {} but input layer is {}".format(
supported_layers, layers)
self.lr_mult_list = lr_mult_list
assert isinstance(self.lr_mult_list, (
list, tuple
)), "lr_mult_list should be in (list, tuple) but got {}".format(
type(self.lr_mult_list))
assert len(
self.lr_mult_list
) == 5, "lr_mult_list length should should be 5 but got {}".format(
len(self.lr_mult_list))
if layers == 18:
depth = [2, 2, 2, 2]
elif layers == 34 or layers == 50:
depth = [3, 4, 6, 3]
elif layers == 101:
depth = [3, 4, 23, 3]
elif layers == 152:
depth = [3, 8, 36, 3]
elif layers == 200:
depth = [3, 12, 48, 3]
num_channels = [64, 256, 512,
1024] if layers >= 50 else [64, 64, 128, 256]
num_filters = [64, 128, 256, 512]
self.conv1_1 = ConvBNLayer(
num_channels=3,
num_filters=32,
filter_size=3,
stride=2,
act='relu',
lr_mult=self.lr_mult_list[0],
name="conv1_1")
self.conv1_2 = ConvBNLayer(
num_channels=32,
num_filters=32,
filter_size=3,
stride=1,
act='relu',
lr_mult=self.lr_mult_list[0],
name="conv1_2")
self.conv1_3 = ConvBNLayer(
num_channels=32,
num_filters=64,
filter_size=3,
stride=1,
act='relu',
lr_mult=self.lr_mult_list[0],
name="conv1_3")
self.pool2d_max = MaxPool2D(kernel_size=3, stride=2, padding=1)
self.block_list = []
if layers >= 50:
for block in range(len(depth)):
shortcut = False
for i in range(depth[block]):
if layers in [101, 152, 200] and block == 2:
if i == 0:
conv_name = "res" + str(block + 2) + "a"
else:
conv_name = "res" + str(block + 2) + "b" + str(i)
else:
conv_name = "res" + str(block + 2) + chr(97 + i)
bottleneck_block = self.add_sublayer(
'bb_%d_%d' % (block, i),
BottleneckBlock(
num_channels=num_channels[block]
if i == 0 else num_filters[block] * 4,
num_filters=num_filters[block],
stride=2 if i == 0 and block != 0 else 1,
shortcut=shortcut,
if_first=block == i == 0,
lr_mult=self.lr_mult_list[block + 1],
name=conv_name))
self.block_list.append(bottleneck_block)
shortcut = True
else:
for block in range(len(depth)):
shortcut = False
for i in range(depth[block]):
conv_name = "res" + str(block + 2) + chr(97 + i)
basic_block = self.add_sublayer(
'bb_%d_%d' % (block, i),
BasicBlock(
num_channels=num_channels[block]
if i == 0 else num_filters[block],
num_filters=num_filters[block],
stride=2 if i == 0 and block != 0 else 1,
shortcut=shortcut,
if_first=block == i == 0,
name=conv_name,
lr_mult=self.lr_mult_list[block + 1]))
self.block_list.append(basic_block)
shortcut = True
self.pool2d_avg = AdaptiveAvgPool2D(1)
self.pool2d_avg_channels = num_channels[-1] * 2
stdv = 1.0 / math.sqrt(self.pool2d_avg_channels * 1.0)
self.out = Linear(
self.pool2d_avg_channels,
class_dim,
weight_attr=ParamAttr(
initializer=Uniform(-stdv, stdv), name="fc_0.w_0"),
bias_attr=ParamAttr(name="fc_0.b_0"))
def forward(self, inputs):
y = self.conv1_1(inputs)
y = self.conv1_2(y)
y = self.conv1_3(y)
y = self.pool2d_max(y)
for block in self.block_list:
y = block(y)
y = self.pool2d_avg(y)
y = paddle.reshape(y, shape=[-1, self.pool2d_avg_channels])
y = self.out(y)
return y
def _load_pretrained(pretrained, model, model_url, use_ssld=False):
if pretrained is False:
pass
elif pretrained is True:
load_dygraph_pretrain_from_url(model, model_url, use_ssld=use_ssld)
elif isinstance(pretrained, str):
load_dygraph_pretrain(model, pretrained)
else:
raise RuntimeError(
"pretrained type is not available. Please use `string` or `boolean` type."
)
def ResNet18_vd(pretrained=False, use_ssld=False, **kwargs):
model = ResNet_vd(layers=18, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet18_vd"], use_ssld=use_ssld)
return model
def ResNet34_vd(pretrained=False, use_ssld=False, **kwargs):
model = ResNet_vd(layers=34, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet34_vd"], use_ssld=use_ssld)
return model
def ResNet50_vd(pretrained=False, use_ssld=False, **kwargs):
model = ResNet_vd(layers=50, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet50_vd"], use_ssld=use_ssld)
return model
def ResNet101_vd(pretrained=False, use_ssld=False, **kwargs):
model = ResNet_vd(layers=101, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet101_vd"], use_ssld=use_ssld)
return model
def ResNet152_vd(pretrained=False, use_ssld=False, **kwargs):
model = ResNet_vd(layers=152, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet152_vd"], use_ssld=use_ssld)
return model
def ResNet200_vd(pretrained=False, use_ssld=False, **kwargs):
model = ResNet_vd(layers=200, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["ResNet200_vd"], use_ssld=use_ssld)
return model
ppcls/arch/backbone/model_zoo/vgg.py 已删除 100644 → 0
浏览文件 @ 3579f5a6
import paddle
from paddle import ParamAttr
import paddle.nn as nn
import paddle.nn.functional as F
from paddle.nn import Conv2D, BatchNorm, Linear, Dropout
from paddle.nn import AdaptiveAvgPool2D, MaxPool2D, AvgPool2D
from ppcls.utils.save_load import load_dygraph_pretrain, load_dygraph_pretrain_from_url
MODEL_URLS = {
"VGG11": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/VGG11_pretrained.pdparams",
"VGG13": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/VGG13_pretrained.pdparams",
"VGG16": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/VGG16_pretrained.pdparams",
"VGG19": "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/VGG19_pretrained.pdparams",
}
__all__ = list(MODEL_URLS.keys())
class ConvBlock(nn.Layer):
def __init__(self, input_channels, output_channels, groups, name=None):
super(ConvBlock, self).__init__()
self.groups = groups
self._conv_1 = Conv2D(
in_channels=input_channels,
out_channels=output_channels,
kernel_size=3,
stride=1,
padding=1,
weight_attr=ParamAttr(name=name + "1_weights"),
bias_attr=False)
if groups == 2 or groups == 3 or groups == 4:
self._conv_2 = Conv2D(
in_channels=output_channels,
out_channels=output_channels,
kernel_size=3,
stride=1,
padding=1,
weight_attr=ParamAttr(name=name + "2_weights"),
bias_attr=False)
if groups == 3 or groups == 4:
self._conv_3 = Conv2D(
in_channels=output_channels,
out_channels=output_channels,
kernel_size=3,
stride=1,
padding=1,
weight_attr=ParamAttr(name=name + "3_weights"),
bias_attr=False)
if groups == 4:
self._conv_4 = Conv2D(
in_channels=output_channels,
out_channels=output_channels,
kernel_size=3,
stride=1,
padding=1,
weight_attr=ParamAttr(name=name + "4_weights"),
bias_attr=False)
self._pool = MaxPool2D(kernel_size=2, stride=2, padding=0)
def forward(self, inputs):
x = self._conv_1(inputs)
x = F.relu(x)
if self.groups == 2 or self.groups == 3 or self.groups == 4:
x = self._conv_2(x)
x = F.relu(x)
if self.groups == 3 or self.groups == 4:
x = self._conv_3(x)
x = F.relu(x)
if self.groups == 4:
x = self._conv_4(x)
x = F.relu(x)
x = self._pool(x)
return x
class VGGNet(nn.Layer):
def __init__(self, layers=11, stop_grad_layers=0, class_dim=1000):
super(VGGNet, self).__init__()
self.layers = layers
self.stop_grad_layers = stop_grad_layers
self.vgg_configure = {
11: [1, 1, 2, 2, 2],
13: [2, 2, 2, 2, 2],
16: [2, 2, 3, 3, 3],
19: [2, 2, 4, 4, 4]
}
assert self.layers in self.vgg_configure.keys(), \
"supported layers are {} but input layer is {}".format(
self.vgg_configure.keys(), layers)
self.groups = self.vgg_configure[self.layers]
self._conv_block_1 = ConvBlock(3, 64, self.groups[0], name="conv1_")
self._conv_block_2 = ConvBlock(64, 128, self.groups[1], name="conv2_")
self._conv_block_3 = ConvBlock(128, 256, self.groups[2], name="conv3_")
self._conv_block_4 = ConvBlock(256, 512, self.groups[3], name="conv4_")
self._conv_block_5 = ConvBlock(512, 512, self.groups[4], name="conv5_")
for idx, block in enumerate([
self._conv_block_1, self._conv_block_2, self._conv_block_3,
self._conv_block_4, self._conv_block_5
]):
if self.stop_grad_layers >= idx + 1:
for param in block.parameters():
param.trainable = False
self._drop = Dropout(p=0.5, mode="downscale_in_infer")
self._fc1 = Linear(
7 * 7 * 512,
4096,
weight_attr=ParamAttr(name="fc6_weights"),
bias_attr=ParamAttr(name="fc6_offset"))
self._fc2 = Linear(
4096,
4096,
weight_attr=ParamAttr(name="fc7_weights"),
bias_attr=ParamAttr(name="fc7_offset"))
self._out = Linear(
4096,
class_dim,
weight_attr=ParamAttr(name="fc8_weights"),
bias_attr=ParamAttr(name="fc8_offset"))
def forward(self, inputs):
x = self._conv_block_1(inputs)
x = self._conv_block_2(x)
x = self._conv_block_3(x)
x = self._conv_block_4(x)
x = self._conv_block_5(x)
x = paddle.flatten(x, start_axis=1, stop_axis=-1)
x = self._fc1(x)
x = F.relu(x)
x = self._drop(x)
x = self._fc2(x)
x = F.relu(x)
x = self._drop(x)
x = self._out(x)
return x
def _load_pretrained(pretrained, model, model_url, use_ssld=False):
if pretrained is False:
pass
elif pretrained is True:
load_dygraph_pretrain_from_url(model, model_url, use_ssld=use_ssld)
elif isinstance(pretrained, str):
load_dygraph_pretrain(model, pretrained)
else:
raise RuntimeError(
"pretrained type is not available. Please use `string` or `boolean` type."
)
def VGG11(pretrained, model, model_url, use_ssld=False, **kwargs):
model = VGGNet(layers=11, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["VGG11"], use_ssld=use_ssld)
return model
def VGG13(pretrained, model, model_url, use_ssld=False, **kwargs):
model = VGGNet(layers=13, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["VGG13"], use_ssld=use_ssld)
return model
def VGG16(pretrained, model, model_url, use_ssld=False, **kwargs):
model = VGGNet(layers=16, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["VGG16"], use_ssld=use_ssld)
return model
def VGG19(pretrained, model, model_url, use_ssld=False, **kwargs):
model = VGGNet(layers=19, **kwargs)
_load_pretrained(pretrained, model, MODEL_URLS["VGG19"], use_ssld=use_ssld)
return model
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