From 8d7f979bf4a0d627a9432f9c0731d06fe2c9ba59 Mon Sep 17 00:00:00 2001 From: anzhengqi Date: Tue, 3 Aug 2021 17:39:39 +0800 Subject: [PATCH] fix pretrained model in psenet network --- model_zoo/official/cv/psenet/README.md | 181 ++++++---- model_zoo/official/cv/psenet/README_CN.md | 207 ++++++----- model_zoo/official/cv/psenet/requirements.txt | 2 + .../cv/psenet/src/ETSNET/pse/Makefile | 3 +- .../cv/psenet/src/ETSNET/pse/adaptor.cpp | 2 +- .../src/psenet_model_torch2mindspore.py | 326 ++++++++++++++++++ model_zoo/official/cv/psenet/train.py | 2 +- 7 files changed, 558 insertions(+), 165 deletions(-) create mode 100644 model_zoo/official/cv/psenet/src/psenet_model_torch2mindspore.py diff --git a/model_zoo/official/cv/psenet/README.md b/model_zoo/official/cv/psenet/README.md index 9e22490b416..a8654fc34b1 100644 --- a/model_zoo/official/cv/psenet/README.md +++ b/model_zoo/official/cv/psenet/README.md @@ -2,6 +2,7 @@ - [PSENet Description](#PSENet-description) - [Dataset](#dataset) +- [Pretrained Model](#Pretrained-model) - [Features](#features) - [Mixed Precision](#mixed-precision) - [Environment Requirements](#environment-requirements) @@ -15,6 +16,7 @@ - [Distributed GPU Training](#distributed-gpu-training) - [Evaluation Process](#evaluation-process) - [Evaluation](#evaluation) + - [Result](#result) - [Inference Process](#inference-process) - [Export MindIR](#export-mindir) - [Infer on Ascend310](#infer-on-ascend310) @@ -48,6 +50,19 @@ Dataset used: [ICDAR2015](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalizatio A training set of 1000 images containing about 4500 readable words A testing set containing about 2000 readable words +unzip dataset files and needn't transform to mindrecord. + +# [Pretrained Model](#contents) + +download pytorch pretrained model: [resnet50-19c8e357.pth](https://download.pytorch.org/models/resnet50-19c8e357.pth) +transform pytorch model to mindspore model + +```shell +cd src + +python psenet_model_torch2mindspore.py --torch_file=/path_to_model/resnet50-19c8e357.pth --output_path=../ +``` + # [Environment Requirements](#contents) - Hardware（Ascend or GPU） @@ -61,34 +76,100 @@ A testing set containing about 2000 readable words - install [pyblind11](https://github.com/pybind/pybind11) - install [Opencv3.4](https://docs.opencv.org/3.4.9/) +```shell +# install pybind11 +pip install pybind11 + +# install opencv3.4.9 +wget https://github.com/opencv/opencv/archive/3.4.9.zip +unzip 3.4.9.zip +cd opencv-3.4.9 +mkdir build +cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local -D WITH_WEBP=OFF .. +make -j4 # -j指定线程数，用户根据机器配置修改参数 +make install + +# export environment variables +export CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/usr/local/include +export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib64 +``` + # [Quick Start](#contents) After installing MindSpore via the official website, you can start training and evaluation as follows: -```python +```shell # run distributed training example bash scripts/run_distribute_train.sh [RANK_TABLE_FILE] [PRED_TRAINED PATH] [TRAIN_ROOT_DIR] -#download opencv library -download pyblind11, opencv3.4 - -#install pyblind11 opencv3.4 -setup pyblind11(install the library by the pip command) -setup opencv3.4(compile source code install the library) - -#enter the path ,run Makefile to product file +#enter the path ,run Makefile cd ./src/ETSNET/pse/;make #run test.py python test.py --ckpt pretrained_model.ckpt --TEST_ROOT_DIR [test root path] -#download eval method from [here](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalization). -#click "My Methods" button,then download Evaluation Scripts +#go to Evaluation Process for details download script.py # run evaluation example bash scripts/run_eval_ascend.sh ``` +- running on ModelArts +- If you want to train the model on modelarts, you can refer to the [official guidance document] of modelarts (https://support.huaweicloud.com/modelarts/) + +```python +# Example of using distributed training on modelarts : +# Data set storage method + +# ├── ICDAR2015 # dir +# ├── train # train dir +# ├── ic15 # train_dataset dir +# ├── ch4_training_images +# ├── ch4_training_localization_transcription_gt +# ├── train_predtrained # predtrained dir +# ├── eval # eval dir +# ├── ic15 # eval dataset dir +# ├── ch4_test_images +# ├── challenge4_Test_Task1_GT +# ├── checkpoint # ckpt files dir + +# (1) Choose either a (modify yaml file parameters) or b (modelArts create training job to modify parameters) 。 +# a. set "enable_modelarts=True" 。 +# set "run_distribute=True" +# set "TRAIN_MODEL_SAVE_PATH=/cache/train/outputs_imagenet/" +# set "TRAIN_ROOT_DIR=/cache/data/ic15/" +# set "pre_trained=/cache/data/train_predtrained/pred file name" Without pre-training weights train_pretrained="" + +# b. add "enable_modelarts=True" Parameters are on the interface of modearts。 +# Set the parameters required by method a on the modelarts interface +# Note: The path parameter does not need to be quoted + +# (2) Set the path of the network configuration file "_config_path=/The path of config in default_config.yaml/" +# (3) Set the code path on the modelarts interface "/path/psenet"。 +# (4) Set the model's startup file on the modelarts interface "train.py" 。 +# (5) Set the data path of the model on the modelarts interface ".../ICDAR2015/train"(choices ICDAR2015/train Folder path) , +# The output path of the model "Output file path" and the log path of the model "Job log path" 。 +# (6) start trainning the model。 + +# Example of using model inference on modelarts +# (1) Place the trained model to the corresponding position of the bucket。 +# (2) chocie a or b。 +# a. set "enable_modelarts=True" 。 +# set "TEST_ROOT_DIR=/cache/data/ic15/" +# set "ckpt=/cache/data/checkpoint/ckpt file" + +# b. Add "enable_modelarts=True" parameter on the interface of modearts。 +# Set the parameters required by method a on the modelarts interface +# Note: The path parameter does not need to be quoted + +# (3) Set the path of the network configuration file "_config_path=/The path of config in default_config.yaml/" +# (4) Set the code path on the modelarts interface "/path/psenet"。 +# (5) Set the model's startup file on the modelarts interface "eval.py" 。 +# (6) Set the data path of the model on the modelarts interface ".../ICDAR2015/eval"(choices ICDAR2015/eval Folder path) , +# The output path of the model "Output file path" and the log path of the model "Job log path" 。 +# (7) Start model inference。 +``` + # [Script Description](#contents) ## [Script and Sample Code](#contents) @@ -156,7 +237,7 @@ Major parameters in default_config.yaml are: Please follow the instructions in the link below: . ```shell -bash scripts/run_distribute_train.sh [RANK_TABLE_FILE] [PRED_TRAINED PATH] [TRAIN_ROOT_DIR] +bash scripts/run_distribute_train.sh [RANK_FILE] [PRETRAINED_PATH] [TRAIN_ROOT_DIR] ``` rank_table_file which is specified by RANK_TABLE_FILE is needed when you are running a distribute task. You can generate it by using the [hccl_tools](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/utils/hccl_tools). @@ -195,66 +276,27 @@ time: 2021-07-24 04:01:07, epoch: 90, step: 31, loss is 0.58495 ### run test code -```test +```shell +python test.py --ckpt [CKPK_PATH] --TEST_ROOT_DIR [TEST_DATA_DIR] + +# click [Here](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalization) to download evaluation scripts +# choose My Methods -> Offline evaluation -> Evaluation Scripts +# download data and put it in /path_to_data +mkdir eval_ic15 +ln -s /path_to_data/script_test_ch4_t1_e1-1577983151.zip eval_ic15/script_test_ch4_t1_e1-1577983151.zip + +cd eval_ic15 +unzip script_test_ch4_t1_e1-1577983151.zip +cd .. + +sh ./script/run_eval_ascend.sh python test.py --ckpt [CKPK PATH] --TEST_ROOT_DIR [TEST DATA DIR] ``` -- running on ModelArts -- If you want to train the model on modelarts, you can refer to the [official guidance document] of modelarts (https://support.huaweicloud.com/modelarts/) +### [Result](#contents) -```python -# Example of using distributed training on modelarts : -# Data set storage method - -# ├── ICDAR2015 # dir -# ├── train # train dir -# ├── ic15 # train_dataset dir -# ├── ch4_training_images -# ├── ch4_training_localization_transcription_gt -# ├── train_predtrained # predtrained dir -# ├── eval # eval dir -# ├── ic15 # eval dataset dir -# ├── ch4_test_images -# ├── challenge4_Test_Task1_GT -# ├── checkpoint # ckpt files dir - -# (1) Choose either a (modify yaml file parameters) or b (modelArts create training job to modify parameters) 。 -# a. set "enable_modelarts=True" 。 -# set "run_distribute=True" -# set "TRAIN_MODEL_SAVE_PATH=/cache/train/outputs_imagenet/" -# set "TRAIN_ROOT_DIR=/cache/data/ic15/" -# set "pre_trained=/cache/data/train_predtrained/pred file name" Without pre-training weights train_pretrained="" - -# b. add "enable_modelarts=True" Parameters are on the interface of modearts。 -# Set the parameters required by method a on the modelarts interface -# Note: The path parameter does not need to be quoted - -# (2) Set the path of the network configuration file "_config_path=/The path of config in default_config.yaml/" -# (3) Set the code path on the modelarts interface "/path/psenet"。 -# (4) Set the model's startup file on the modelarts interface "train.py" 。 -# (5) Set the data path of the model on the modelarts interface ".../ICDAR2015/train"(choices ICDAR2015/train Folder path) , -# The output path of the model "Output file path" and the log path of the model "Job log path" 。 -# (6) start trainning the model。 - -# Example of using model inference on modelarts -# (1) Place the trained model to the corresponding position of the bucket。 -# (2) chocie a or b。 -# a. set "enable_modelarts=True" 。 -# set "TEST_ROOT_DIR=/cache/data/ic15/" -# set "ckpt=/cache/data/checkpoint/ckpt file" - -# b. Add "enable_modelarts=True" parameter on the interface of modearts。 -# Set the parameters required by method a on the modelarts interface -# Note: The path parameter does not need to be quoted - -# (3) Set the path of the network configuration file "_config_path=/The path of config in default_config.yaml/" -# (4) Set the code path on the modelarts interface "/path/psenet"。 -# (5) Set the model's startup file on the modelarts interface "eval.py" 。 -# (6) Set the data path of the model on the modelarts interface ".../ICDAR2015/eval"(choices ICDAR2015/eval Folder path) , -# The output path of the model "Output file path" and the log path of the model "Job log path" 。 -# (7) Start model inference。 -``` +Calculated!{"precision": 0.8147966668299853，"recall"：0.8006740491092923，"hmean"：0.8076736279747451，"AP"：0} ### Eval Script for ICDAR2015 @@ -342,8 +384,9 @@ The `res` folder is generated in the upper-level directory. For details about th | Loss Function | LossCallBack | | outputs | probability | | Loss | 0.35 | -| Speed | 1pc: 444 ms/step; 8pcs: 446 ms/step | -| Total time | 1pc: 75.48 h; 8pcs: 7.11 h | +| Parameters | batch_size = 4 | +| Speed | 1pc: 444 ms/step(fps: 9.0); 8pcs: 446 ms/step(fps: 71) | +| Total time | 1pc: 75.48 h; 8pcs: 7.11 h | | Parameters (M) | 27.36 | | Checkpoint for Fine tuning | 109.44M (.ckpt file) | | Scripts | | diff --git a/model_zoo/official/cv/psenet/README_CN.md b/model_zoo/official/cv/psenet/README_CN.md index 7355e1e44e8..770e9872bac 100644 --- a/model_zoo/official/cv/psenet/README_CN.md +++ b/model_zoo/official/cv/psenet/README_CN.md @@ -5,6 +5,7 @@ - [PSENet示例](#psenet示例) - [概述](#概述) - [数据集](#数据集) +- [预训练模型](#预训练模型) - [环境要求](#环境要求) - [快速入门](#快速入门) - [脚本说明](#脚本说明) @@ -14,9 +15,7 @@ - [分布式训练](#分布式训练) - [评估过程](#评估过程) - [运行测试代码](#运行测试代码) - - [ICDAR2015评估脚本](#icdar2015评估脚本) - - [用法](#用法) - - [结果](#结果) + - [结果](#结果) - [推理过程](#推理过程) - [导出MindIR](#导出mindir) - [在Ascend310执行推理](#在ascend310执行推理) @@ -48,6 +47,21 @@ 训练集：包括约4500个可读单词的1000张图像。 测试集：约2000个可读单词。 +下载得到的训练和推理数据解压后备用，不需要转为mindrecord数据 + +# 预训练模型 + +下载pytorch的预训练模型: [resnet50-19c8e357.pth](https://download.pytorch.org/models/resnet50-19c8e357.pth) +将pytorch模型转为mindspore模型 + +```shell +cd src + +python psenet_model_torch2mindspore.py --torch_file=/path_to_model/resnet50-19c8e357.pth --output_path=../ +``` + +执行完成，src的上层目录得到文件pretrained_model.ckpt文件，用于接下来的训练 + # 环境要求 - 硬件：昇腾处理器（Ascend） @@ -62,36 +76,100 @@ - 安装[pyblind11](https://github.com/pybind/pybind11) - 安装[Opencv3.4](https://docs.opencv.org/3.4.9/) +```shell +# 使用pip安装pybind11 +pip install pybind11 + +# 使用源码安装opencv3.4.9 +wget https://github.com/opencv/opencv/archive/3.4.9.zip +unzip 3.4.9.zip +cd opencv-3.4.9 +mkdir build +cmake -D CMAKE_BUILD_TYPE=Release -D CMAKE_INSTALL_PREFIX=/usr/local -D WITH_WEBP=OFF .. +make -j4 # -j指定线程数，用户根据机器配置修改参数 +make install + +# opencv安装在/usr/local目录下，将该目录添加到环境变量中 +export CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/usr/local/include +export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib64 +``` + # 快速入门 通过官方网站安装MindSpore后，您可以按照如下步骤进行训练和评估： -```python +```shell # 分布式训练运行示例 -bash scripts/run_distribute_train.sh [RANK_TABLE_FILE] [PRED_TRAINED PATH] [TRAIN_ROOT_DIR] +# 第一个参数为rank_table文件，第二个参数为生成的预训练模型，第三个参数为下载的训练数据集 +bash scripts/run_distribute_train.sh [RANK_FILE] [PRETRAINED_PATH] [TRAIN_ROOT_DIR] -# 下载opencv库 -download pyblind11, opencv3.4 - -# 安装pyblind11 opencv3.4 -setup pyblind11(install the library by the pip command) -setup opencv3.4(compile source code install the library) - -# 单击[此处](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalization)下载评估方法 -# 点击"我的方法"按钮，下载评估脚本 - -# 输入路径，运行Makefile，找到产品文件 +# 进入路径，运行Makefile cd ./src/ETSNET/pse/;make clean&&make # 运行test.py -python test.py --ckpt pretrained_model.ckpt --TEST_ROOT_DIR [test root path] - +python test.py --ckpt [CKPK_PATH] --TEST_ROOT_DIR [TEST_DATA_DIR] +# 具体见评估过程 download script.py # 运行评估示例 bash scripts/run_eval_ascend.sh ``` +- 如果要在modelarts上进行模型的训练，可以参考modelarts的[官方指导文档](https://support.huaweicloud.com/modelarts/) 开始进行模型的训练和推理，具体操作如下： + +```ModelArts +# 在ModelArts上使用分布式训练示例: +# 数据集存放方式 + +# ├── ICDAR2015 # dir +# ├── train # train dir +# ├── ic15 # train_dataset dir +# ├── ch4_training_images +# ├── ch4_training_localization_transcription_gt +# ├── train_predtrained # predtrained dir +# ├── eval # eval dir +# ├── ic15 # eval dataset dir +# ├── ch4_test_images +# ├── challenge4_Test_Task1_GT +# ├── checkpoint # ckpt files dir + +# (1) 选择a(修改yaml文件参数)或者b(ModelArts创建训练作业修改参数)其中一种方式。 +# a. 设置 "enable_modelarts=True" +# 设置 "run_distribute=True" +# 设置 "TRAIN_MODEL_SAVE_PATH=/cache/train/outputs/" +# 设置 "TRAIN_ROOT_DIR=/cache/data/ic15/" +# 设置 "pre_trained=/cache/data/train_predtrained/pred file name" 如果没有预训练权重 pre_trained="" + +# b. 增加 "enable_modelarts=True" 参数在modearts的界面上。 +# 在modelarts的界面上设置方法a所需要的参数 +# 注意：路径参数不需要加引号 + +# (2)设置网络配置文件的路径 "_config_path=/The path of config in default_config.yaml/" +# (3) 在modelarts的界面上设置代码的路径 "/path/psenet"。 +# (4) 在modelarts的界面上设置模型的启动文件 "train.py" 。 +# (5) 在modelarts的界面上设置模型的数据路径 ".../ICDAR2015/train"(选择ICDAR2015/train文件夹路径) , +# 模型的输出路径"Output file path" 和模型的日志路径 "Job log path" 。 +# (6) 开始模型的训练。 + +# 在modelarts上使用模型推理的示例 +# (1) 把训练好的模型地方到桶的对应位置。 +# (2) 选择a或者b其中一种方式。 +# a.设置 "enable_modelarts=True" +# 设置 "TEST_ROOT_DIR=/cache/data/ic15" +# 设置 "ckpt=/cache/data/checkpoint/ckpt file" + +# b. 增加 "enable_modelarts=True" 参数在modearts的界面上。 +# 在modelarts的界面上设置方法a所需要的参数 +# 注意：路径参数不需要加引号 + +# (3) 设置网络配置文件的路径 "_config_path=/The path of config in default_config.yaml/" +# (4) 在modelarts的界面上设置代码的路径 "/path/psenet"。 +# (5) 在modelarts的界面上设置模型的启动文件 "eval.py" 。 +# (6) 在modelarts的界面上设置模型的数据路径 "../ICDAR2015/eval"(选择ICDAR2015/eval文件夹路径) , +# 模型的输出路径"Output file path" 和模型的日志路径 "Job log path" 。 +# (7) 开始模型的推理。 +``` + ## 脚本说明 ## 脚本和样例代码 @@ -153,7 +231,8 @@ bash scripts/run_eval_ascend.sh 请遵循链接中的说明：[链接](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/utils/hccl_tools) ```shell -bash scripts/run_distribute_train.sh [RANK_TABLE_FILE] [PRED_TRAINED PATH] [TRAIN_ROOT_DIR] +# 第一个参数为rank_table文件，第二个参数为生成的预训练模型，第三个参数为下载的训练数据集 +bash scripts/run_distribute_train.sh [RANK_FILE] [PRETRAINED_PATH] [TRAIN_ROOT_DIR] ``` 上述shell脚本将在后台运行分布训练。可以通过`device[X]/test_*.log`文件查看结果。 @@ -173,81 +252,24 @@ device_1/log:epcoh： 2, step: 40，loss is 0.76629 ### 运行测试代码 -```test -python test.py --ckpt [CKPK PATH] --TEST_ROOT_DIR [TEST DATA DIR] - -``` - -- 如果要在modelarts上进行模型的训练，可以参考modelarts的[官方指导文档](https://support.huaweicloud.com/modelarts/) 开始进行模型的训练和推理，具体操作如下： - -```ModelArts -# 在ModelArts上使用分布式训练示例: -# 数据集存放方式 - -# ├── ICDAR2015 # dir -# ├── train # train dir -# ├── ic15 # train_dataset dir -# ├── ch4_training_images -# ├── ch4_training_localization_transcription_gt -# ├── train_predtrained # predtrained dir -# ├── eval # eval dir -# ├── ic15 # eval dataset dir -# ├── ch4_test_images -# ├── challenge4_Test_Task1_GT -# ├── checkpoint # ckpt files dir - -# (1) 选择a(修改yaml文件参数)或者b(ModelArts创建训练作业修改参数)其中一种方式。 -# a. 设置 "enable_modelarts=True" -# 设置 "run_distribute=True" -# 设置 "TRAIN_MODEL_SAVE_PATH=/cache/train/outputs/" -# 设置 "TRAIN_ROOT_DIR=/cache/data/ic15/" -# 设置 "pre_trained=/cache/data/train_predtrained/pred file name" 如果没有预训练权重 pre_trained="" - -# b. 增加 "enable_modelarts=True" 参数在modearts的界面上。 -# 在modelarts的界面上设置方法a所需要的参数 -# 注意：路径参数不需要加引号 - -# (2)设置网络配置文件的路径 "_config_path=/The path of config in default_config.yaml/" -# (3) 在modelarts的界面上设置代码的路径 "/path/psenet"。 -# (4) 在modelarts的界面上设置模型的启动文件 "train.py" 。 -# (5) 在modelarts的界面上设置模型的数据路径 ".../ICDAR2015/train"(选择ICDAR2015/train文件夹路径) , -# 模型的输出路径"Output file path" 和模型的日志路径 "Job log path" 。 -# (6) 开始模型的训练。 - -# 在modelarts上使用模型推理的示例 -# (1) 把训练好的模型地方到桶的对应位置。 -# (2) 选择a或者b其中一种方式。 -# a.设置 "enable_modelarts=True" -# 设置 "TEST_ROOT_DIR=/cache/data/ic15" -# 设置 "ckpt=/cache/data/checkpoint/ckpt file" - -# b. 增加 "enable_modelarts=True" 参数在modearts的界面上。 -# 在modelarts的界面上设置方法a所需要的参数 -# 注意：路径参数不需要加引号 - -# (3) 设置网络配置文件的路径 "_config_path=/The path of config in default_config.yaml/" -# (4) 在modelarts的界面上设置代码的路径 "/path/psenet"。 -# (5) 在modelarts的界面上设置模型的启动文件 "eval.py" 。 -# (6) 在modelarts的界面上设置模型的数据路径 "../ICDAR2015/eval"(选择ICDAR2015/eval文件夹路径) , -# 模型的输出路径"Output file path" 和模型的日志路径 "Job log path" 。 -# (7) 开始模型的推理。 -``` - -### ICDAR2015评估脚本 - -#### 用法 - -第一步：单击[此处](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalization)下载评估方法。 - -第二步：单击"我的方法"按钮，下载评估脚本。 - -第三步：建议将评估方法根符号链接到$MINDSPORE/model_zoo/psenet/eval_ic15/。如果您的文件夹结构不同，您可能需要更改评估脚本文件中的相应路径。 - ```shell -bash ./script/run_eval_ascend.sh.sh +# 第一个参数为训练得到的模型文件，第二个参数为下载得到的推理数据集 +python test.py --ckpt [CKPK_PATH] --TEST_ROOT_DIR [TEST_DATA_DIR] + +# 单击[此处](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalization)下载评估方法 +# 点击"My Methods"按钮，选择Offline evaluation -> Evaluation Scripts +# 下载完成后，将数据放在/path_to_data路径 +mkdir eval_ic15 +ln -s /path_to_data/script_test_ch4_t1_e1-1577983151.zip eval_ic15/script_test_ch4_t1_e1-1577983151.zip + +cd eval_ic15 +unzip script_test_ch4_t1_e1-1577983151.zip +cd .. + +bash ./script/run_eval_ascend.sh ``` -#### 结果 +### 结果 Calculated!{"precision": 0.8147966668299853，"recall"：0.8006740491092923，"hmean"：0.8076736279747451，"AP"：0} @@ -317,7 +339,8 @@ bash run_infer_310.sh [MINDIR_PATH] [DATA_PATH] [DEVICE_ID] | 损失函数 | LossCallBack | | 输出 | 概率 | | 损失 | 0.35 | -| 速度 | 1卡：444毫秒/步；8卡：446毫秒/步 +| 训练参数 | batch_size = 4 | +| 速度 | 1卡：444毫秒/步(fps: 9.0)；8卡：446毫秒/步(fps: 71) | | 总时间 | 1卡：75.48小时；8卡：7.11小时| | 参数(M) | 27.36 | | 微调检查点 | 109.44M （.ckpt file） | diff --git a/model_zoo/official/cv/psenet/requirements.txt b/model_zoo/official/cv/psenet/requirements.txt index 9d316731512..bee48e58af9 100644 --- a/model_zoo/official/cv/psenet/requirements.txt +++ b/model_zoo/official/cv/psenet/requirements.txt @@ -2,3 +2,5 @@ numpy opencv-python pillow pyyaml +Polygon3 +pyclipper diff --git a/model_zoo/official/cv/psenet/src/ETSNET/pse/Makefile b/model_zoo/official/cv/psenet/src/ETSNET/pse/Makefile index 541e9ba3b37..eac5bc1e8e6 100644 --- a/model_zoo/official/cv/psenet/src/ETSNET/pse/Makefile +++ b/model_zoo/official/cv/psenet/src/ETSNET/pse/Makefile @@ -13,8 +13,7 @@ # limitations under the License. # ============================================================================ -mindspore_home = ${MINDSPORE_HOME} -CXXFLAGS = -I include -I ${mindspore_home}/model_zoo/official/cv/psenet -std=c++11 -O3 +CXXFLAGS = -std=c++11 -O3 CXX_SOURCES = adaptor.cpp opencv_home = ${OPENCV_HOME} OPENCV = -I$(opencv_home)/include -L$(opencv_home)/lib64 -lopencv_superres -lopencv_ml -lopencv_objdetect \ diff --git a/model_zoo/official/cv/psenet/src/ETSNET/pse/adaptor.cpp b/model_zoo/official/cv/psenet/src/ETSNET/pse/adaptor.cpp index 8885e848fec..f4e343e9fc3 100644 --- a/model_zoo/official/cv/psenet/src/ETSNET/pse/adaptor.cpp +++ b/model_zoo/official/cv/psenet/src/ETSNET/pse/adaptor.cpp @@ -13,7 +13,6 @@ * See the License for the specific language governing permissions and * limitations under the License. */ -#include "src/ETSNET/pse/adaptor.h" #include #include #include @@ -26,6 +25,7 @@ #include #include #include +#include "./adaptor.h" using std::vector; using std::queue; diff --git a/model_zoo/official/cv/psenet/src/psenet_model_torch2mindspore.py b/model_zoo/official/cv/psenet/src/psenet_model_torch2mindspore.py new file mode 100644 index 00000000000..a5a0d6fb892 --- /dev/null +++ b/model_zoo/official/cv/psenet/src/psenet_model_torch2mindspore.py @@ -0,0 +1,326 @@ +# Copyright 2021 Huawei Technologies Co., Ltd +# +# 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. +# ============================================================================ +"""transform pytorch checkpoint to mindspore checkpoint""" +import os +import argparse + +import torch +from mindspore import Tensor +from mindspore.train.serialization import save_checkpoint + +parser = argparse.ArgumentParser(description="transform pytorch checkpoint to mindspore checkpoint") +parser.add_argument("--torch_file", type=str, required=True, help="input pytorch checkpoint filename") +parser.add_argument("--output_path", type=str, required=True, help="output mindspore checkpoint path") +args = parser.parse_args() + +weights_dict = { + "feature_extractor.conv1.weight": "conv1.weight", + "feature_extractor.bn1.moving_mean": "bn1.running_mean", + "feature_extractor.bn1.moving_variance": "bn1.running_var", + "feature_extractor.bn1.gamma": "bn1.weight", + "feature_extractor.bn1.beta": "bn1.bias", + + "feature_extractor.layer1.0.conv1.weight": "layer1.0.conv1.weight", + "feature_extractor.layer1.0.bn1.moving_mean": "layer1.0.bn1.running_mean", + "feature_extractor.layer1.0.bn1.moving_variance": "layer1.0.bn1.running_var", + "feature_extractor.layer1.0.bn1.gamma": "layer1.0.bn1.weight", + "feature_extractor.layer1.0.bn1.beta": "layer1.0.bn1.bias", + "feature_extractor.layer1.0.conv2.weight": "layer1.0.conv2.weight", + "feature_extractor.layer1.0.bn2.moving_mean": "layer1.0.bn2.running_mean", + "feature_extractor.layer1.0.bn2.moving_variance": "layer1.0.bn2.running_var", + "feature_extractor.layer1.0.bn2.gamma": "layer1.0.bn2.weight", + "feature_extractor.layer1.0.bn2.beta": "layer1.0.bn2.bias", + "feature_extractor.layer1.0.conv3.weight": "layer1.0.conv3.weight", + "feature_extractor.layer1.0.bn3.moving_mean": "layer1.0.bn3.running_mean", + "feature_extractor.layer1.0.bn3.moving_variance": "layer1.0.bn3.running_var", + "feature_extractor.layer1.0.bn3.gamma": "layer1.0.bn3.weight", + "feature_extractor.layer1.0.bn3.beta": "layer1.0.bn3.bias", + "feature_extractor.layer1.0.conv_down_sample.weight": "layer1.0.downsample.0.weight", + "feature_extractor.layer1.0.bn_down_sample.moving_mean": "layer1.0.downsample.1.running_mean", + "feature_extractor.layer1.0.bn_down_sample.moving_variance": "layer1.0.downsample.1.running_var", + "feature_extractor.layer1.0.bn_down_sample.gamma": "layer1.0.downsample.1.weight", + "feature_extractor.layer1.0.bn_down_sample.beta": "layer1.0.downsample.1.bias", + + "feature_extractor.layer1.1.conv1.weight": "layer1.1.conv1.weight", + "feature_extractor.layer1.1.bn1.moving_mean": "layer1.1.bn1.running_mean", + "feature_extractor.layer1.1.bn1.moving_variance": "layer1.1.bn1.running_var", + "feature_extractor.layer1.1.bn1.gamma": "layer1.1.bn1.weight", + "feature_extractor.layer1.1.bn1.beta": "layer1.1.bn1.bias", + "feature_extractor.layer1.1.conv2.weight": "layer1.1.conv2.weight", + "feature_extractor.layer1.1.bn2.moving_mean": "layer1.1.bn2.running_mean", + "feature_extractor.layer1.1.bn2.moving_variance": "layer1.1.bn2.running_var", + "feature_extractor.layer1.1.bn2.gamma": "layer1.1.bn2.weight", + "feature_extractor.layer1.1.bn2.beta": "layer1.1.bn2.bias", + "feature_extractor.layer1.1.conv3.weight": "layer1.1.conv3.weight", + "feature_extractor.layer1.1.bn3.moving_mean": "layer1.1.bn3.running_mean", + "feature_extractor.layer1.1.bn3.moving_variance": "layer1.1.bn3.running_var", + "feature_extractor.layer1.1.bn3.gamma": "layer1.1.bn3.weight", + "feature_extractor.layer1.1.bn3.beta": "layer1.1.bn3.bias", + + "feature_extractor.layer1.2.conv1.weight": "layer1.2.conv1.weight", + "feature_extractor.layer1.2.bn1.moving_mean": "layer1.2.bn1.running_mean", + "feature_extractor.layer1.2.bn1.moving_variance": "layer1.2.bn1.running_var", + "feature_extractor.layer1.2.bn1.gamma": "layer1.2.bn1.weight", + "feature_extractor.layer1.2.bn1.beta": "layer1.2.bn1.bias", + "feature_extractor.layer1.2.conv2.weight": "layer1.2.conv2.weight", + "feature_extractor.layer1.2.bn2.moving_mean": "layer1.2.bn2.running_mean", + "feature_extractor.layer1.2.bn2.moving_variance": "layer1.2.bn2.running_var", + "feature_extractor.layer1.2.bn2.gamma": "layer1.2.bn2.weight", + "feature_extractor.layer1.2.bn2.beta": "layer1.2.bn2.bias", + "feature_extractor.layer1.2.conv3.weight": "layer1.2.conv3.weight", + "feature_extractor.layer1.2.bn3.moving_mean": "layer1.2.bn3.running_mean", + "feature_extractor.layer1.2.bn3.moving_variance": "layer1.2.bn3.running_var", + "feature_extractor.layer1.2.bn3.gamma": "layer1.2.bn3.weight", + "feature_extractor.layer1.2.bn3.beta": "layer1.2.bn3.bias", + + "feature_extractor.layer2.0.conv1.weight": "layer2.0.conv1.weight", + "feature_extractor.layer2.0.bn1.moving_mean": "layer2.0.bn1.running_mean", + "feature_extractor.layer2.0.bn1.moving_variance": "layer2.0.bn1.running_var", + "feature_extractor.layer2.0.bn1.gamma": "layer2.0.bn1.weight", + "feature_extractor.layer2.0.bn1.beta": "layer2.0.bn1.bias", + "feature_extractor.layer2.0.conv2.weight": "layer2.0.conv2.weight", + "feature_extractor.layer2.0.bn2.moving_mean": "layer2.0.bn2.running_mean", + "feature_extractor.layer2.0.bn2.moving_variance": "layer2.0.bn2.running_var", + "feature_extractor.layer2.0.bn2.gamma": "layer2.0.bn2.weight", + "feature_extractor.layer2.0.bn2.beta": "layer2.0.bn2.bias", + "feature_extractor.layer2.0.conv3.weight": "layer2.0.conv3.weight", + "feature_extractor.layer2.0.bn3.moving_mean": "layer2.0.bn3.running_mean", + "feature_extractor.layer2.0.bn3.moving_variance": "layer2.0.bn3.running_var", + "feature_extractor.layer2.0.bn3.gamma": "layer2.0.bn3.weight", + "feature_extractor.layer2.0.bn3.beta": "layer2.0.bn3.bias", + "feature_extractor.layer2.0.conv_down_sample.weight": "layer2.0.downsample.0.weight", + "feature_extractor.layer2.0.bn_down_sample.moving_mean": "layer2.0.downsample.1.running_mean", + "feature_extractor.layer2.0.bn_down_sample.moving_variance": "layer2.0.downsample.1.running_var", + "feature_extractor.layer2.0.bn_down_sample.gamma": "layer2.0.downsample.1.weight", + "feature_extractor.layer2.0.bn_down_sample.beta": "layer2.0.downsample.1.bias", + + "feature_extractor.layer2.1.conv1.weight": "layer2.1.conv1.weight", + "feature_extractor.layer2.1.bn1.moving_mean": "layer2.1.bn1.running_mean", + "feature_extractor.layer2.1.bn1.moving_variance": "layer2.1.bn1.running_var", + "feature_extractor.layer2.1.bn1.gamma": "layer2.1.bn1.weight", + "feature_extractor.layer2.1.bn1.beta": "layer2.1.bn1.bias", + "feature_extractor.layer2.1.conv2.weight": "layer2.1.conv2.weight", + "feature_extractor.layer2.1.bn2.moving_mean": "layer2.1.bn2.running_mean", + "feature_extractor.layer2.1.bn2.moving_variance": "layer2.1.bn2.running_var", + "feature_extractor.layer2.1.bn2.gamma": "layer2.1.bn2.weight", + "feature_extractor.layer2.1.bn2.beta": "layer2.1.bn2.bias", + "feature_extractor.layer2.1.conv3.weight": "layer2.1.conv3.weight", + "feature_extractor.layer2.1.bn3.moving_mean": "layer2.1.bn3.running_mean", + "feature_extractor.layer2.1.bn3.moving_variance": "layer2.1.bn3.running_var", + "feature_extractor.layer2.1.bn3.gamma": "layer2.1.bn3.weight", + "feature_extractor.layer2.1.bn3.beta": "layer2.1.bn3.bias", + + "feature_extractor.layer2.2.conv1.weight": "layer2.2.conv1.weight", + "feature_extractor.layer2.2.bn1.moving_mean": "layer2.2.bn1.running_mean", + "feature_extractor.layer2.2.bn1.moving_variance": "layer2.2.bn1.running_var", + "feature_extractor.layer2.2.bn1.gamma": "layer2.2.bn1.weight", + "feature_extractor.layer2.2.bn1.beta": "layer2.2.bn1.bias", + "feature_extractor.layer2.2.conv2.weight": "layer2.2.conv2.weight", + "feature_extractor.layer2.2.bn2.moving_mean": "layer2.2.bn2.running_mean", + "feature_extractor.layer2.2.bn2.moving_variance": "layer2.2.bn2.running_var", + "feature_extractor.layer2.2.bn2.gamma": "layer2.2.bn2.weight", + "feature_extractor.layer2.2.bn2.beta": "layer2.2.bn2.bias", + "feature_extractor.layer2.2.conv3.weight": "layer2.2.conv3.weight", + "feature_extractor.layer2.2.bn3.moving_mean": "layer2.2.bn3.running_mean", + "feature_extractor.layer2.2.bn3.moving_variance": "layer2.2.bn3.running_var", + "feature_extractor.layer2.2.bn3.gamma": "layer2.2.bn3.weight", + "feature_extractor.layer2.2.bn3.beta": "layer2.2.bn3.bias", + + "feature_extractor.layer2.3.conv1.weight": "layer2.3.conv1.weight", + "feature_extractor.layer2.3.bn1.moving_mean": "layer2.3.bn1.running_mean", + "feature_extractor.layer2.3.bn1.moving_variance": "layer2.3.bn1.running_var", + "feature_extractor.layer2.3.bn1.gamma": "layer2.3.bn1.weight", + "feature_extractor.layer2.3.bn1.beta": "layer2.3.bn1.bias", + "feature_extractor.layer2.3.conv2.weight": "layer2.3.conv2.weight", + "feature_extractor.layer2.3.bn2.moving_mean": "layer2.3.bn2.running_mean", + "feature_extractor.layer2.3.bn2.moving_variance": "layer2.3.bn2.running_var", + "feature_extractor.layer2.3.bn2.gamma": "layer2.3.bn2.weight", + "feature_extractor.layer2.3.bn2.beta": "layer2.3.bn2.bias", + "feature_extractor.layer2.3.conv3.weight": "layer2.3.conv3.weight", + "feature_extractor.layer2.3.bn3.moving_mean": "layer2.3.bn3.running_mean", + "feature_extractor.layer2.3.bn3.moving_variance": "layer2.3.bn3.running_var", + "feature_extractor.layer2.3.bn3.gamma": "layer2.3.bn3.weight", + "feature_extractor.layer2.3.bn3.beta": "layer2.3.bn3.bias", + + "feature_extractor.layer3.0.conv1.weight": "layer3.0.conv1.weight", + "feature_extractor.layer3.0.bn1.moving_mean": "layer3.0.bn1.running_mean", + "feature_extractor.layer3.0.bn1.moving_variance": "layer3.0.bn1.running_var", + "feature_extractor.layer3.0.bn1.gamma": "layer3.0.bn1.weight", + "feature_extractor.layer3.0.bn1.beta": "layer3.0.bn1.bias", + "feature_extractor.layer3.0.conv2.weight": "layer3.0.conv2.weight", + "feature_extractor.layer3.0.bn2.moving_mean": "layer3.0.bn2.running_mean", + "feature_extractor.layer3.0.bn2.moving_variance": "layer3.0.bn2.running_var", + "feature_extractor.layer3.0.bn2.gamma": "layer3.0.bn2.weight", + "feature_extractor.layer3.0.bn2.beta": "layer3.0.bn2.bias", + "feature_extractor.layer3.0.conv3.weight": "layer3.0.conv3.weight", + "feature_extractor.layer3.0.bn3.moving_mean": "layer3.0.bn3.running_mean", + "feature_extractor.layer3.0.bn3.moving_variance": "layer3.0.bn3.running_var", + "feature_extractor.layer3.0.bn3.gamma": "layer3.0.bn3.weight", + "feature_extractor.layer3.0.bn3.beta": "layer3.0.bn3.bias", + "feature_extractor.layer3.0.conv_down_sample.weight": "layer3.0.downsample.0.weight", + "feature_extractor.layer3.0.bn_down_sample.moving_mean": "layer3.0.downsample.1.running_mean", + "feature_extractor.layer3.0.bn_down_sample.moving_variance": "layer3.0.downsample.1.running_var", + "feature_extractor.layer3.0.bn_down_sample.gamma": "layer3.0.downsample.1.weight", + "feature_extractor.layer3.0.bn_down_sample.beta": "layer3.0.downsample.1.bias", + + "feature_extractor.layer3.1.conv1.weight": "layer3.1.conv1.weight", + "feature_extractor.layer3.1.bn1.moving_mean": "layer3.1.bn1.running_mean", + "feature_extractor.layer3.1.bn1.moving_variance": "layer3.1.bn1.running_var", + "feature_extractor.layer3.1.bn1.gamma": "layer3.1.bn1.weight", + "feature_extractor.layer3.1.bn1.beta": "layer3.1.bn1.bias", + "feature_extractor.layer3.1.conv2.weight": "layer3.1.conv2.weight", + "feature_extractor.layer3.1.bn2.moving_mean": "layer3.1.bn2.running_mean", + "feature_extractor.layer3.1.bn2.moving_variance": "layer3.1.bn2.running_var", + "feature_extractor.layer3.1.bn2.gamma": "layer3.1.bn2.weight", + "feature_extractor.layer3.1.bn2.beta": "layer3.1.bn2.bias", + "feature_extractor.layer3.1.conv3.weight": "layer3.1.conv3.weight", + "feature_extractor.layer3.1.bn3.moving_mean": "layer3.1.bn3.running_mean", + "feature_extractor.layer3.1.bn3.moving_variance": "layer3.1.bn3.running_var", + "feature_extractor.layer3.1.bn3.gamma": "layer3.1.bn3.weight", + "feature_extractor.layer3.1.bn3.beta": "layer3.1.bn3.bias", + + "feature_extractor.layer3.2.conv1.weight": "layer3.2.conv1.weight", + "feature_extractor.layer3.2.bn1.moving_mean": "layer3.2.bn1.running_mean", + "feature_extractor.layer3.2.bn1.moving_variance": "layer3.2.bn1.running_var", + "feature_extractor.layer3.2.bn1.gamma": "layer3.2.bn1.weight", + "feature_extractor.layer3.2.bn1.beta": "layer3.2.bn1.bias", + "feature_extractor.layer3.2.conv2.weight": "layer3.2.conv2.weight", + "feature_extractor.layer3.2.bn2.moving_mean": "layer3.2.bn2.running_mean", + "feature_extractor.layer3.2.bn2.moving_variance": "layer3.2.bn2.running_var", + "feature_extractor.layer3.2.bn2.gamma": "layer3.2.bn2.weight", + "feature_extractor.layer3.2.bn2.beta": "layer3.2.bn2.bias", + "feature_extractor.layer3.2.conv3.weight": "layer3.2.conv3.weight", + "feature_extractor.layer3.2.bn3.moving_mean": "layer3.2.bn3.running_mean", + "feature_extractor.layer3.2.bn3.moving_variance": "layer3.2.bn3.running_var", + "feature_extractor.layer3.2.bn3.gamma": "layer3.2.bn3.weight", + "feature_extractor.layer3.2.bn3.beta": "layer3.2.bn3.bias", + + "feature_extractor.layer3.3.conv1.weight": "layer3.3.conv1.weight", + "feature_extractor.layer3.3.bn1.moving_mean": "layer3.3.bn1.running_mean", + "feature_extractor.layer3.3.bn1.moving_variance": "layer3.3.bn1.running_var", + "feature_extractor.layer3.3.bn1.gamma": "layer3.3.bn1.weight", + "feature_extractor.layer3.3.bn1.beta": "layer3.3.bn1.bias", + "feature_extractor.layer3.3.conv2.weight": "layer3.3.conv2.weight", + "feature_extractor.layer3.3.bn2.moving_mean": "layer3.3.bn2.running_mean", + "feature_extractor.layer3.3.bn2.moving_variance": "layer3.3.bn2.running_var", + "feature_extractor.layer3.3.bn2.gamma": "layer3.3.bn2.weight", + "feature_extractor.layer3.3.bn2.beta": "layer3.3.bn2.bias", + "feature_extractor.layer3.3.conv3.weight": "layer3.3.conv3.weight", + "feature_extractor.layer3.3.bn3.moving_mean": "layer3.3.bn3.running_mean", + "feature_extractor.layer3.3.bn3.moving_variance": "layer3.3.bn3.running_var", + "feature_extractor.layer3.3.bn3.gamma": "layer3.3.bn3.weight", + "feature_extractor.layer3.3.bn3.beta": "layer3.3.bn3.bias", + + "feature_extractor.layer3.4.conv1.weight": "layer3.4.conv1.weight", + "feature_extractor.layer3.4.bn1.moving_mean": "layer3.4.bn1.running_mean", + "feature_extractor.layer3.4.bn1.moving_variance": "layer3.4.bn1.running_var", + "feature_extractor.layer3.4.bn1.gamma": "layer3.4.bn1.weight", + "feature_extractor.layer3.4.bn1.beta": "layer3.4.bn1.bias", + "feature_extractor.layer3.4.conv2.weight": "layer3.4.conv2.weight", + "feature_extractor.layer3.4.bn2.moving_mean": "layer3.4.bn2.running_mean", + "feature_extractor.layer3.4.bn2.moving_variance": "layer3.4.bn2.running_var", + "feature_extractor.layer3.4.bn2.gamma": "layer3.4.bn2.weight", + "feature_extractor.layer3.4.bn2.beta": "layer3.4.bn2.bias", + "feature_extractor.layer3.4.conv3.weight": "layer3.4.conv3.weight", + "feature_extractor.layer3.4.bn3.moving_mean": "layer3.4.bn3.running_mean", + "feature_extractor.layer3.4.bn3.moving_variance": "layer3.4.bn3.running_var", + "feature_extractor.layer3.4.bn3.gamma": "layer3.4.bn3.weight", + "feature_extractor.layer3.4.bn3.beta": "layer3.4.bn3.bias", + + "feature_extractor.layer3.5.conv1.weight": "layer3.5.conv1.weight", + "feature_extractor.layer3.5.bn1.moving_mean": "layer3.5.bn1.running_mean", + "feature_extractor.layer3.5.bn1.moving_variance": "layer3.5.bn1.running_var", + "feature_extractor.layer3.5.bn1.gamma": "layer3.5.bn1.weight", + "feature_extractor.layer3.5.bn1.beta": "layer3.5.bn1.bias", + "feature_extractor.layer3.5.conv2.weight": "layer3.5.conv2.weight", + "feature_extractor.layer3.5.bn2.moving_mean": "layer3.5.bn2.running_mean", + "feature_extractor.layer3.5.bn2.moving_variance": "layer3.5.bn2.running_var", + "feature_extractor.layer3.5.bn2.gamma": "layer3.5.bn2.weight", + "feature_extractor.layer3.5.bn2.beta": "layer3.5.bn2.bias", + "feature_extractor.layer3.5.conv3.weight": "layer3.5.conv3.weight", + "feature_extractor.layer3.5.bn3.moving_mean": "layer3.5.bn3.running_mean", + "feature_extractor.layer3.5.bn3.moving_variance": "layer3.5.bn3.running_var", + "feature_extractor.layer3.5.bn3.gamma": "layer3.5.bn3.weight", + "feature_extractor.layer3.5.bn3.beta": "layer3.5.bn3.bias", + + "feature_extractor.layer4.0.conv1.weight": "layer4.0.conv1.weight", + "feature_extractor.layer4.0.bn1.moving_mean": "layer4.0.bn1.running_mean", + "feature_extractor.layer4.0.bn1.moving_variance": "layer4.0.bn1.running_var", + "feature_extractor.layer4.0.bn1.gamma": "layer4.0.bn1.weight", + "feature_extractor.layer4.0.bn1.beta": "layer4.0.bn1.bias", + "feature_extractor.layer4.0.conv2.weight": "layer4.0.conv2.weight", + "feature_extractor.layer4.0.bn2.moving_mean": "layer4.0.bn2.running_mean", + "feature_extractor.layer4.0.bn2.moving_variance": "layer4.0.bn2.running_var", + "feature_extractor.layer4.0.bn2.gamma": "layer4.0.bn2.weight", + "feature_extractor.layer4.0.bn2.beta": "layer4.0.bn2.bias", + "feature_extractor.layer4.0.conv3.weight": "layer4.0.conv3.weight", + "feature_extractor.layer4.0.bn3.moving_mean": "layer4.0.bn3.running_mean", + "feature_extractor.layer4.0.bn3.moving_variance": "layer4.0.bn3.running_var", + "feature_extractor.layer4.0.bn3.gamma": "layer4.0.bn3.weight", + "feature_extractor.layer4.0.bn3.beta": "layer4.0.bn3.bias", + "feature_extractor.layer4.0.conv_down_sample.weight": "layer4.0.downsample.0.weight", + "feature_extractor.layer4.0.bn_down_sample.moving_mean": "layer4.0.downsample.1.running_mean", + "feature_extractor.layer4.0.bn_down_sample.moving_variance": "layer4.0.downsample.1.running_var", + "feature_extractor.layer4.0.bn_down_sample.gamma": "layer4.0.downsample.1.weight", + "feature_extractor.layer4.0.bn_down_sample.beta": "layer4.0.downsample.1.bias", + + "feature_extractor.layer4.1.conv1.weight": "layer4.1.conv1.weight", + "feature_extractor.layer4.1.bn1.moving_mean": "layer4.1.bn1.running_mean", + "feature_extractor.layer4.1.bn1.moving_variance": "layer4.1.bn1.running_var", + "feature_extractor.layer4.1.bn1.gamma": "layer4.1.bn1.weight", + "feature_extractor.layer4.1.bn1.beta": "layer4.1.bn1.bias", + "feature_extractor.layer4.1.conv2.weight": "layer4.1.conv2.weight", + "feature_extractor.layer4.1.bn2.moving_mean": "layer4.1.bn2.running_mean", + "feature_extractor.layer4.1.bn2.moving_variance": "layer4.1.bn2.running_var", + "feature_extractor.layer4.1.bn2.gamma": "layer4.1.bn2.weight", + "feature_extractor.layer4.1.bn2.beta": "layer4.1.bn2.bias", + "feature_extractor.layer4.1.conv3.weight": "layer4.1.conv3.weight", + "feature_extractor.layer4.1.bn3.moving_mean": "layer4.1.bn3.running_mean", + "feature_extractor.layer4.1.bn3.moving_variance": "layer4.1.bn3.running_var", + "feature_extractor.layer4.1.bn3.gamma": "layer4.1.bn3.weight", + "feature_extractor.layer4.1.bn3.beta": "layer4.1.bn3.bias", + + "feature_extractor.layer4.2.conv1.weight": "layer4.2.conv1.weight", + "feature_extractor.layer4.2.bn1.moving_mean": "layer4.2.bn1.running_mean", + "feature_extractor.layer4.2.bn1.moving_variance": "layer4.2.bn1.running_var", + "feature_extractor.layer4.2.bn1.gamma": "layer4.2.bn1.weight", + "feature_extractor.layer4.2.bn1.beta": "layer4.2.bn1.bias", + "feature_extractor.layer4.2.conv2.weight": "layer4.2.conv2.weight", + "feature_extractor.layer4.2.bn2.moving_mean": "layer4.2.bn2.running_mean", + "feature_extractor.layer4.2.bn2.moving_variance": "layer4.2.bn2.running_var", + "feature_extractor.layer4.2.bn2.gamma": "layer4.2.bn2.weight", + "feature_extractor.layer4.2.bn2.beta": "layer4.2.bn2.bias", + "feature_extractor.layer4.2.conv3.weight": "layer4.2.conv3.weight", + "feature_extractor.layer4.2.bn3.moving_mean": "layer4.2.bn3.running_mean", + "feature_extractor.layer4.2.bn3.moving_variance": "layer4.2.bn3.running_var", + "feature_extractor.layer4.2.bn3.gamma": "layer4.2.bn3.weight", + "feature_extractor.layer4.2.bn3.beta": "layer4.2.bn3.bias", +} + +torch_param_dict = torch.load(args.torch_file) +ms_params = [] + +for ms_name, torch_name in weights_dict.items(): + if torch_name not in torch_param_dict.keys(): + print("param {} not in pytorch checkpoint".format(torch_name)) + continue + torch_value = torch_param_dict[torch_name] + np_value = torch_value.data.numpy() + each_param = dict() + each_param["name"] = ms_name + each_param["data"] = Tensor(np_value) + ms_params.append(each_param) + +save_checkpoint(ms_params, os.path.join(args.output_path, "pretrained_model.ckpt")) diff --git a/model_zoo/official/cv/psenet/train.py b/model_zoo/official/cv/psenet/train.py index b11e45ecce5..d8519e2fd51 100644 --- a/model_zoo/official/cv/psenet/train.py +++ b/model_zoo/official/cv/psenet/train.py @@ -100,7 +100,7 @@ def train(): if config.pre_trained: param_dict = load_checkpoint(config.pre_trained) - load_param_into_net(net, param_dict) + load_param_into_net(net, param_dict, strict_load=True) print('Load Pretrained parameters done!') criterion = DiceLoss(batch_size=config.TRAIN_BATCH_SIZE)