提交 16572f27 编写于 作者: L LDOUBLEV

add doc of decommpressing pretrain weight file

上级 90d2bf0c
...@@ -43,6 +43,18 @@ cd PaddleOCR/ ...@@ -43,6 +43,18 @@ cd PaddleOCR/
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV3_large_x0_5_pretrained.tar wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV3_large_x0_5_pretrained.tar
# 下载ResNet50的预训练模型 # 下载ResNet50的预训练模型
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_ssld_pretrained.tar wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_ssld_pretrained.tar
# 解压预训练模型文件,以MobileNetV3为例
tar xf ./pretrain_models/MobileNetV3_large_x0_5_pretrained.tar ./pretrain_models/
# 注:正确解压backbone预训练权重文件后,文件夹下包含众多以网络层命名的权重文件,格式如下:
./pretrain_models/MobileNetV3_large_x0_5_pretrained/
└─ conv_last_bn_mean
└─ conv_last_bn_offset
└─ conv_last_bn_scale
└─ conv_last_bn_variance
└─ ......
``` ```
**启动训练** **启动训练**
......
...@@ -42,6 +42,18 @@ cd PaddleOCR/ ...@@ -42,6 +42,18 @@ cd PaddleOCR/
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV3_large_x0_5_pretrained.tar wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/MobileNetV3_large_x0_5_pretrained.tar
# Download the pre-trained model of ResNet50 # Download the pre-trained model of ResNet50
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_ssld_pretrained.tar wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_ssld_pretrained.tar
# decompressing the pre-training model file, take MobileNetV3 as an example
tar xf ./pretrain_models/MobileNetV3_large_x0_5_pretrained.tar ./pretrain_models/
# Note: After decompressing the backbone pre-training weight file correctly, the file list in the folder is as follows:
./pretrain_models/MobileNetV3_large_x0_5_pretrained/
└─ conv_last_bn_mean
└─ conv_last_bn_offset
└─ conv_last_bn_scale
└─ conv_last_bn_variance
└─ ......
``` ```
**Start training** **Start training**
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册