Merge branch 'PaddlePaddle:dygraph' into dygraph

test_tipc/prepare_lite.sh

-#!/bin/bash
-source ./test_tipc/common_func.sh
-FILENAME=$1
-dataline=$(cat ${FILENAME})
-# parser params
-IFS=$'\n'
-lines=(${dataline})
-IFS=$'\n'
-lite_model_list=$(func_parser_value "${lines[2]}")
-
-# prepare lite .nb model
-pip install paddlelite==2.9
-current_dir=${PWD}
-IFS="|"
-model_path=./inference_models
-for model in ${lite_model_list[*]}; do
-    inference_model_url=https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/${model}.tar
-    inference_model=${inference_model_url##*/}
-    wget -nc  -P ${model_path} ${inference_model_url}
-    cd ${model_path} && tar -xf ${inference_model} && cd ../
-    model_dir=${model_path}/${inference_model%.*}
-    model_file=${model_dir}/inference.pdmodel
-    param_file=${model_dir}/inference.pdiparams
-    paddle_lite_opt --model_dir=${model_dir} --model_file=${model_file} --param_file=${param_file} --valid_targets=arm --optimize_out=${model_dir}_opt
-done
-
-# prepare test data
-data_url=https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/icdar2015_lite.tar
-model_path=./inference_models
-inference_model=${inference_model_url##*/}
-data_file=${data_url##*/}
-wget -nc  -P ./inference_models ${inference_model_url}
-wget -nc  -P ./test_data ${data_url}
-cd ./inference_models && tar -xf ${inference_model} && cd ../
-cd ./test_data && tar -xf ${data_file} && rm ${data_file} && cd ../
-
-# prepare lite env
-paddlelite_url=https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.9/inference_lite_lib.android.armv8.gcc.c++_shared.with_extra.with_cv.tar.gz
-paddlelite_zipfile=$(echo $paddlelite_url | awk -F "/" '{print $NF}')
-paddlelite_file=${paddlelite_zipfile:0:66}
-wget ${paddlelite_url} && tar -xf ${paddlelite_zipfile}
-mkdir -p  ${paddlelite_file}/demo/cxx/ocr/test_lite
-cp -r ${model_path}/*_opt.nb test_data ${paddlelite_file}/demo/cxx/ocr/test_lite
-cp ppocr/utils/ppocr_keys_v1.txt deploy/lite/config.txt ${paddlelite_file}/demo/cxx/ocr/test_lite
-cp -r ./deploy/lite/* ${paddlelite_file}/demo/cxx/ocr/
-cp ${paddlelite_file}/cxx/lib/libpaddle_light_api_shared.so ${paddlelite_file}/demo/cxx/ocr/test_lite
-cp ${FILENAME} test_tipc/test_lite_arm_cpu_cpp.sh test_tipc/common_func.sh ${paddlelite_file}/demo/cxx/ocr/test_lite
-cd ${paddlelite_file}/demo/cxx/ocr/
-git clone https://github.com/cuicheng01/AutoLog.git
-make -j
-sleep 1
-make -j
-cp ocr_db_crnn test_lite && cp test_lite/libpaddle_light_api_shared.so test_lite/libc++_shared.so
-tar -cf test_lite.tar ./test_lite && cp test_lite.tar ${current_dir} && cd ${current_dir}
-rm -rf ${paddlelite_file}* && rm -rf ${model_path}

test_tipc/prepare_lite_cpp.sh

+#!/bin/bash
+source ./test_tipc/common_func.sh
+FILENAME=$1
+dataline=$(cat ${FILENAME})
+# parser params
+IFS=$'\n'
+lines=(${dataline})
+IFS=$'\n'
+paddlelite_library_source=$2
+
+inference_cmd=$(func_parser_value "${lines[1]}")
+DEVICE=$(func_parser_value "${lines[2]}")
+det_lite_model_list=$(func_parser_value "${lines[3]}")
+rec_lite_model_list=$(func_parser_value "${lines[4]}")
+cls_lite_model_list=$(func_parser_value "${lines[5]}")
+
+if [[ $inference_cmd =~ "det" ]]; then
+    lite_model_list=${det_lite_model_list}
+elif [[ $inference_cmd =~ "rec" ]]; then
+    lite_model_list=(${rec_lite_model_list[*]} ${cls_lite_model_list[*]})
+elif [[ $inference_cmd =~ "system" ]]; then
+    lite_model_list=(${det_lite_model_list[*]} ${rec_lite_model_list[*]} ${cls_lite_model_list[*]})
+else
+    echo "inference_cmd is wrong, please check."
+    exit 1
+fi
+
+if [ ${DEVICE} = "ARM_CPU" ]; then
+    valid_targets="arm"
+    paddlelite_library_url="https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.10-rc/inference_lite_lib.android.armv8.gcc.c++_shared.with_extra.with_cv.tar.gz"
+    end_index="66"
+    compile_with_opencl="OFF"
+elif [ ${DEVICE} = "ARM_GPU_OPENCL" ]; then
+    valid_targets="opencl"
+    paddlelite_library_url="https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.10-rc/inference_lite_lib.armv8.clang.with_exception.with_extra.with_cv.opencl.tar.gz"
+    end_index="71"
+    compile_with_opencl="ON"
+else
+    echo "DEVICE only support ARM_CPU, ARM_GPU_OPENCL."
+    exit 2    
+fi
+
+# prepare paddlelite model
+pip install paddlelite==2.10-rc
+current_dir=${PWD}
+IFS="|"
+model_path=./inference_models
+
+for model in ${lite_model_list[*]}; do
+    if [[ $model =~ "PP-OCRv2" ]]; then
+        inference_model_url=https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/${model}.tar
+    elif [[ $model =~ "v2.0" ]]; then
+        inference_model_url=https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/${model}.tar
+    else 
+        echo "Model is wrong, please check."
+        exit 3
+    fi
+    inference_model=${inference_model_url##*/}
+    wget -nc  -P ${model_path} ${inference_model_url}
+    cd ${model_path} && tar -xf ${inference_model} && cd ../
+    model_dir=${model_path}/${inference_model%.*}
+    model_file=${model_dir}/inference.pdmodel
+    param_file=${model_dir}/inference.pdiparams
+    paddle_lite_opt --model_dir=${model_dir} --model_file=${model_file} --param_file=${param_file} --valid_targets=${valid_targets} --optimize_out=${model_dir}_opt
+done
+
+# prepare test data
+data_url=https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/icdar2015_lite.tar
+data_file=${data_url##*/}
+wget -nc  -P ./test_data ${data_url}
+cd ./test_data && tar -xf ${data_file} && rm ${data_file} && cd ../
+
+# prepare paddlelite predict library
+if [[ ${paddlelite_library_source} = "download" ]]; then
+    paddlelite_library_zipfile=$(echo $paddlelite_library_url | awk -F "/" '{print $NF}')
+    paddlelite_library_file=${paddlelite_library_zipfile:0:${end_index}}
+    wget ${paddlelite_library_url} && tar -xf ${paddlelite_library_zipfile}
+    cd ${paddlelite_library_zipfile}
+elif [[ ${paddlelite_library_source} = "compile" ]]; then
+    git clone -b release/v2.10 https://github.com/PaddlePaddle/Paddle-Lite.git
+    cd Paddle-Lite
+    ./lite/tools/build_android.sh  --arch=armv8  --with_cv=ON --with_extra=ON --toolchain=clang --with_opencl=${compile_with_opencl}
+    cd ../
+    cp -r Paddle-Lite/build.lite.android.armv8.clang/inference_lite_lib.android.armv8/ .
+    paddlelite_library_file=inference_lite_lib.android.armv8
+else
+    echo "paddlelite_library_source only support 'download' and 'compile'"
+    exit 3
+fi
+
+# organize the required files  
+mkdir -p  ${paddlelite_library_file}/demo/cxx/ocr/test_lite
+cp -r ${model_path}/*_opt.nb test_data ${paddlelite_library_file}/demo/cxx/ocr/test_lite
+cp ppocr/utils/ppocr_keys_v1.txt deploy/lite/config.txt ${paddlelite_library_file}/demo/cxx/ocr/test_lite
+cp -r ./deploy/lite/* ${paddlelite_library_file}/demo/cxx/ocr/
+cp ${paddlelite_library_file}/cxx/lib/libpaddle_light_api_shared.so ${paddlelite_library_file}/demo/cxx/ocr/test_lite
+cp ${FILENAME} test_tipc/test_lite_arm_cpp.sh test_tipc/common_func.sh ${paddlelite_library_file}/demo/cxx/ocr/test_lite
+cd ${paddlelite_library_file}/demo/cxx/ocr/
+git clone https://github.com/cuicheng01/AutoLog.git
+
+# compile and do some postprocess
+make -j
+sleep 1
+make -j
+cp ocr_db_crnn test_lite && cp test_lite/libpaddle_light_api_shared.so test_lite/libc++_shared.so
+tar -cf test_lite.tar ./test_lite && cp test_lite.tar ${current_dir} && cd ${current_dir}
+rm -rf ${paddlelite_library_file}* && rm -rf ${model_path}

test_tipc/readme.md

 
-# 飞桨训推一体认证
+# 飞桨训推一体全流程（TIPC）
 
 ## 1. 简介
 
-飞桨除了基本的模型训练和预测，还提供了支持多端多平台的高性能推理部署工具。本文档提供了PaddleOCR中所有模型的飞桨训推一体认证 (Training and Inference Pipeline Certification(TIPC)) 信息和测试工具，方便用户查阅每种模型的训练推理部署打通情况，并可以进行一键测试。
+飞桨除了基本的模型训练和预测，还提供了支持多端多平台的高性能推理部署工具。本文档提供了PaddleOCR中所有模型的飞桨训推一体全流程（Training and Inference Pipeline Criterion(TIPC)）信息和测试工具，方便用户查阅每种模型的训练推理部署打通情况，并可以进行一键测试。
 
 <div align="center">
     <img src="docs/guide.png" width="1000">
@@ -23,14 +23,17 @@
 
 | 算法论文 | 模型名称 | 模型类型 | 基础<br>训练预测 | 更多<br>训练方式 | 模型压缩 |  其他预测部署  |
 | :--- | :--- |  :----:  | :--------: |  :----  |   :----  |   :----  |
-| DB     |ch_ppocr_mobile_v2.0_det | 检测  | 支持 | 多机多卡 <br> 混合精度 | FPGM裁剪 <br> 离线量化| Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
-| DB     |ch_ppocr_server_v2.0_det | 检测  | 支持 | 多机多卡 <br> 混合精度 | FPGM裁剪 <br> 离线量化| Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
+| DB     |ch_ppocr_mobile_v2.0_det | 检测  | 支持 | 多机多卡 <br> 混合精度 | - | Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
+| DB     |ch_ppocr_mobile_v2.0_det_FPGM | 检测  | 支持 | 多机多卡 <br> 混合精度 | FPGM裁剪 | Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
+| DB     |ch_ppocr_mobile_v2.0_det_PACT | 检测  | 支持 | 多机多卡 <br> 混合精度 | PACT量化 | Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
+| DB     |ch_ppocr_mobile_v2.0_det_KL | 检测  | 支持 | 多机多卡 <br> 混合精度 | 离线量化| Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
+| DB     |ch_ppocr_server_v2.0_det | 检测  | 支持 | 多机多卡 <br> 混合精度 | - | Paddle Inference: C++ <br> Paddle Serving: Python, C++ |
 | DB     |ch_PP-OCRv2_det          | 检测  |
-| CRNN   |ch_ppocr_mobile_v2.0_rec | 识别  | 支持 | 多机多卡 <br> 混合精度 | PACT量化 <br> 离线量化| Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
-| CRNN   |ch_ppocr_server_v2.0_rec | 识别  | 支持 | 多机多卡 <br> 混合精度 | PACT量化 <br> 离线量化| Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
+| CRNN   |ch_ppocr_mobile_v2.0_rec | 识别  | 支持 | 多机多卡 <br> 混合精度 | - | Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
+| CRNN   |ch_ppocr_server_v2.0_rec | 识别  | 支持 | 多机多卡 <br> 混合精度 | - | Paddle Inference: C++ <br> Paddle Serving: Python, C++ |
 | CRNN   |ch_PP-OCRv2_rec          | 识别  |
 | PP-OCR |ch_ppocr_mobile_v2.0 | 检测+识别  | 支持 | 多机多卡 <br> 混合精度 | - | Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
-| PP-OCR |ch_ppocr_server_v2.0 | 检测+识别  | 支持 | 多机多卡 <br> 混合精度 | - | Paddle Inference: C++ <br> Paddle Serving: Python, C++ <br> Paddle-Lite: <br> (1) ARM CPU(C++) |
+| PP-OCR |ch_ppocr_server_v2.0 | 检测+识别  | 支持 | 多机多卡 <br> 混合精度 | - | Paddle Inference: C++ <br> Paddle Serving: Python, C++ |
 |PP-OCRv2|ch_PP-OCRv2 | 检测+识别  |
 | DB     |det_mv3_db_v2.0                | 检测  |
 | DB     |det_r50_vd_db_v2.0             | 检测  |
@@ -54,32 +57,31 @@
 
 
 
-## 3. 一键测试工具使用
+## 3. 测试工具简介
 ### 目录介绍
 
 ```shell
 test_tipc/
 ├── configs/  # 配置文件目录
-    ├── ppocr_det_mobile             # ppocr_det_mobile模型的测试配置文件目录
-        ├── det_mv3_db.yml                  # 测试mobile版ppocr检测模型训练的yml文件
-        ├── train_infer_python.txt.txt      # 测试Linux上python训练预测（基础训练预测）的配置文件
+    ├── ch_ppocr_mobile_v2.0_det    # ch_ppocr_mobile_v2.0_det模型的测试配置文件目录
+        ├── train_infer_python.txt      # 测试Linux上python训练预测（基础训练预测）的配置文件
         ├── model_linux_gpu_normal_normal_infer_cpp_linux_gpu_cpu.txt     # 测试Linux上c++预测的配置文件
         ├── model_linux_gpu_normal_normal_infer_python_jetson.txt         # 测试Jetson上python预测的配置文件
         ├── train_linux_gpu_fleet_amp_infer_python_linux_gpu_cpu.txt      # 测试Linux上多机多卡、混合精度训练和python预测的配置文件
         ├── ...  
-    ├── ppocr_det_server               # ppocr_det_server模型的测试配置文件目录
+    ├── ch_ppocr_server_v2.0_det               # ch_ppocr_server_v2.0_det模型的测试配置文件目录
         ├── ...  
-    ├── ppocr_rec_mobile               # ppocr_rec_mobile模型的测试配置文件目录
+    ├── ch_ppocr_mobile_v2.0_rec               # ch_ppocr_mobile_v2.0_rec模型的测试配置文件目录
         ├── ...  
-    ├── ppocr_rec_server               # ppocr_rec_server模型的测试配置文件目录
+    ├── ch_ppocr_server_v2.0_det               # ch_ppocr_server_v2.0_det模型的测试配置文件目录
         ├── ...  
     ├── ...  
 ├── results/   # 预先保存的预测结果，用于和实际预测结果进行精读比对
-	├── python_ppocr_det_mobile_results_fp32.txt           # 预存的mobile版ppocr检测模型python预测fp32精度的结果
-	├── python_ppocr_det_mobile_results_fp16.txt           # 预存的mobile版ppocr检测模型python预测fp16精度的结果
-	├── cpp_ppocr_det_mobile_results_fp32.txt       # 预存的mobile版ppocr检测模型c++预测的fp32精度的结果
-	├── cpp_ppocr_det_mobile_results_fp16.txt       # 预存的mobile版ppocr检测模型c++预测的fp16精度的结果
-	├── ...
+    ├── python_ppocr_det_mobile_results_fp32.txt           # 预存的mobile版ppocr检测模型python预测fp32精度的结果
+    ├── python_ppocr_det_mobile_results_fp16.txt           # 预存的mobile版ppocr检测模型python预测fp16精度的结果
+    ├── cpp_ppocr_det_mobile_results_fp32.txt       # 预存的mobile版ppocr检测模型c++预测的fp32精度的结果
+    ├── cpp_ppocr_det_mobile_results_fp16.txt       # 预存的mobile版ppocr检测模型c++预测的fp16精度的结果
+    ├── ...
 ├── prepare.sh                        # 完成test_*.sh运行所需要的数据和模型下载
 ├── test_train_inference_python.sh    # 测试python训练预测的主程序
 ├── test_inference_cpp.sh             # 测试c++预测的主程序
@@ -89,19 +91,9 @@ test_tipc/
 └── readme.md                         # 使用文档
 ```
 
-### 配置文件命名规范
-在`configs`目录下，按模型名称划分为子目录，子目录中存放所有该模型测试需要用到的配置文件，配置文件的命名遵循如下规范：
-
-1. 基础训练预测配置简单命名为：`train_infer_python.txt`，表示**Linux环境下单机、不使用混合精度训练+python预测**，其完整命名对应`train_linux_gpu_fleet_amp_infer_python_linux_gpu_cpu.txt`，由于本配置文件使用频率较高，这里进行了名称简化。
-
-2. 其他带训练配置命名格式为：`train_训练硬件环境(linux_gpu/linux_dcu/…)_是否多机(fleet/normal)_是否混合精度(amp/normal)_预测模式(infer/lite/serving/js)_语言(cpp/python/java)_预测硬件环境(linux_gpu/mac/jetson/opencl_arm_gpu/...).txt`。如，linux gpu下多机多卡+混合精度链条测试对应配置 `train_linux_gpu_fleet_amp_infer_python_linux_gpu_cpu.txt`，linux dcu下基础训练预测对应配置 `train_linux_dcu_normal_normal_infer_python_dcu.txt`。
-
-3. 仅预测的配置（如serving、lite等）命名格式：`model_训练硬件环境(linux_gpu/linux_dcu/…)_是否多机(fleet/normal)_是否混合精度(amp/normal)_(infer/lite/serving/js)_语言(cpp/python/java)_预测硬件环境(linux_gpu/mac/jetson/opencl_arm_gpu/...).txt`，即，与2相比，仅第一个字段从train换为model，测试时模型直接下载获取，这里的“训练硬件环境”表示所测试的模型是在哪种环境下训练得到的。
+### 测试流程概述
 
-根据上述命名规范，可以直接从配置文件名看出对应的测试场景和功能。
-
-### 测试流程
-使用本工具，可以测试不同功能的支持情况，以及预测结果是否对齐，测试流程如下：
+使用本工具，可以测试不同功能的支持情况，以及预测结果是否对齐，测试流程概括如下：
 <div align="center">
     <img src="docs/test.png" width="800">
 </div>
@@ -110,18 +102,43 @@ test_tipc/
 2. 运行要测试的功能对应的测试脚本`test_*.sh`，产出log，由log可以看到不同配置是否运行成功；
 3. 用`compare_results.py`对比log中的预测结果和预存在results目录下的结果，判断预测精度是否符合预期（在误差范围内）。
 
-其中，有4个测试主程序，功能如下：
-- `test_train_inference_python.sh`：测试基于Python的模型训练、评估、推理等基本功能，包括裁剪、量化、蒸馏。
-- `test_inference_cpp.sh`：测试基于C++的模型推理。
-- `test_serving.sh`：测试基于Paddle Serving的服务化部署功能。
-- `test_lite_arm_cpu_cpp.sh`：测试基于Paddle-Lite的ARM CPU端c++预测部署功能。
-- `test_paddle2onnx.sh`：测试Paddle2ONNX的模型转化功能，并验证正确性。
+测试单项功能仅需两行命令，**如需测试不同模型/功能，替换配置文件即可**，命令格式如下：
+```shell
+# 功能：准备数据
+# 格式：bash + 运行脚本 + 参数1: 配置文件选择 + 参数2: 模式选择
+bash test_tipc/prepare.sh  configs/[model_name]/[params_file_name]  [Mode]
+
+# 功能：运行测试
+# 格式：bash + 运行脚本 + 参数1: 配置文件选择 + 参数2: 模式选择
+bash test_tipc/test_train_inference_python.sh configs/[model_name]/[params_file_name]  [Mode]
+```
+
+例如，测试基本训练预测功能的`lite_train_lite_infer`模式，运行：
+```shell
+# 准备数据
+bash test_tipc/prepare.sh ./test_tipc/configs/ch_ppocr_mobile_v2.0_det/train_infer_python.txt 'lite_train_lite_infer'
+# 运行测试
+bash test_tipc/test_train_inference_python.sh ./test_tipc/configs/ch_ppocr_mobile_v2.0_det/train_infer_python.txt 'lite_train_lite_infer'
+```  
+关于本示例命令的更多信息可查看[基础训练预测使用文档](https://github.com/PaddlePaddle/PaddleOCR/blob/dygraph/test_tipc/docs/test_train_inference_python.md#22-%E5%8A%9F%E8%83%BD%E6%B5%8B%E8%AF%95)。
+
+### 配置文件命名规范
+在`configs`目录下，**按模型名称划分为子目录**，子目录中存放所有该模型测试需要用到的配置文件，配置文件的命名遵循如下规范：
+
+1. 基础训练预测配置简单命名为：`train_infer_python.txt`，表示**Linux环境下单机、不使用混合精度训练+python预测**，其完整命名对应`train_linux_gpu_normal_normal_infer_python_linux_gpu_cpu.txt`，由于本配置文件使用频率较高，这里进行了名称简化。
+
+2. 其他带训练配置命名格式为：`train_训练硬件环境(linux_gpu/linux_dcu/…)_是否多机(fleet/normal)_是否混合精度(amp/normal)_预测模式(infer/lite/serving/js)_语言(cpp/python/java)_预测硬件环境(linux_gpu/mac/jetson/opencl_arm_gpu/...).txt`。如，linux gpu下多机多卡+混合精度链条测试对应配置 `train_linux_gpu_fleet_amp_infer_python_linux_gpu_cpu.txt`，linux dcu下基础训练预测对应配置 `train_linux_dcu_normal_normal_infer_python_linux_dcu.txt`。
+
+3. 仅预测的配置（如serving、lite等）命名格式：`model_训练硬件环境(linux_gpu/linux_dcu/…)_是否多机(fleet/normal)_是否混合精度(amp/normal)_(infer/lite/serving/js)_语言(cpp/python/java)_预测硬件环境(linux_gpu/mac/jetson/opencl_arm_gpu/...).txt`，即，与2相比，仅第一个字段从train换为model，测试时模型直接下载获取，这里的“训练硬件环境”表示所测试的模型是在哪种环境下训练得到的。
+
+**根据上述命名规范，可以直接从子目录名称和配置文件名找到需要测试的场景和功能对应的配置文件。**
+
 
 <a name="more"></a>
-#### 更多教程
+## 4. 开始测试
 各功能测试中涉及混合精度、裁剪、量化等训练相关，及mkldnn、Tensorrt等多种预测相关参数配置，请点击下方相应链接了解更多细节和使用教程：  
-[test_train_inference_python 使用](docs/test_train_inference_python.md)  
-[test_inference_cpp 使用](docs/test_inference_cpp.md)  
-[test_serving 使用](docs/test_serving.md)  
-[test_lite_arm_cpu_cpp 使用](docs/test_lite_arm_cpu_cpp.md)  
-[test_paddle2onnx 使用](docs/test_paddle2onnx.md)  
+- [test_train_inference_python 使用](docs/test_train_inference_python.md) ：测试基于Python的模型训练、评估、推理等基本功能，包括裁剪、量化、蒸馏。 
+- [test_inference_cpp 使用](docs/test_inference_cpp.md)：测试基于C++的模型推理。
+- [test_serving 使用](docs/test_serving.md)：测试基于Paddle Serving的服务化部署功能。
+- [test_lite_arm_cpu_cpp 使用](docs/test_lite_arm_cpu_cpp.md)：测试基于Paddle-Lite的ARM CPU端c++预测部署功能。
+- [test_paddle2onnx 使用](docs/test_paddle2onnx.md)：测试Paddle2ONNX的模型转化功能，并验证正确性。

test_tipc/test_inference_jeston.sh → test_tipc/test_inference_python.sh

 #!/bin/bash
 source test_tipc/common_func.sh
-source test_tipc/test_train_inference_python.sh
+#source test_tipc/test_train_inference_python.sh
 
 FILENAME=$1
 # MODE be one of ['whole_infer']
 MODE=$2
 
-dataline=$(awk 'NR==1, NR==17{print}'  $FILENAME)
+dataline=$(awk 'NR==1, NR==20{print}'  $FILENAME)
 
 # parser params
 IFS=$'\n'
@@ -35,18 +35,100 @@ precision_list=$(func_parser_value "${lines[12]}")
 infer_model_key=$(func_parser_key "${lines[13]}")
 image_dir_key=$(func_parser_key "${lines[14]}")
 infer_img_dir=$(func_parser_value "${lines[14]}")
-save_log_key=$(func_parser_key "${lines[15]}")
+rec_model_key=$(func_parser_key "${lines[15]}")
+rec_model_value=$(func_parser_value "${lines[15]}")
 benchmark_key=$(func_parser_key "${lines[16]}")
 benchmark_value=$(func_parser_value "${lines[16]}")
 infer_key1=$(func_parser_key "${lines[17]}")
 infer_value1=$(func_parser_value "${lines[17]}")
 
 
+
 LOG_PATH="./test_tipc/output"
 mkdir -p ${LOG_PATH}
 status_log="${LOG_PATH}/results_python.log"
 
 
+function func_inference(){
+    IFS='|'
+    _python=$1
+    _script=$2
+    _model_dir=$3
+    _log_path=$4
+    _img_dir=$5
+    _flag_quant=$6
+    # inference 
+    for use_gpu in ${use_gpu_list[*]}; do
+        if [ ${use_gpu} = "False" ] || [ ${use_gpu} = "cpu" ]; then
+            for use_mkldnn in ${use_mkldnn_list[*]}; do
+                if [ ${use_mkldnn} = "False" ] && [ ${_flag_quant} = "True" ]; then
+                    continue
+                fi
+                for threads in ${cpu_threads_list[*]}; do
+                    for batch_size in ${batch_size_list[*]}; do
+                        for precision in ${precision_list[*]}; do
+                            if [ ${use_mkldnn} = "False" ] && [ ${precision} = "fp16" ]; then
+                                continue
+                            fi # skip when enable fp16 but disable mkldnn
+                            if [ ${_flag_quant} = "True" ] && [ ${precision} != "int8" ]; then
+                                continue
+                            fi # skip when quant model inference but precision is not int8
+                            set_precision=$(func_set_params "${precision_key}" "${precision}")
+                            
+                            _save_log_path="${_log_path}/python_infer_cpu_usemkldnn_${use_mkldnn}_threads_${threads}_precision_${precision}_batchsize_${batch_size}.log"
+                            set_infer_data=$(func_set_params "${image_dir_key}" "${_img_dir}")
+                            set_benchmark=$(func_set_params "${benchmark_key}" "${benchmark_value}")
+                            set_batchsize=$(func_set_params "${batch_size_key}" "${batch_size}")
+                            set_cpu_threads=$(func_set_params "${cpu_threads_key}" "${threads}")
+                            set_model_dir=$(func_set_params "${infer_model_key}" "${_model_dir}")
+                            set_infer_params0=$(func_set_params "${rec_model_key}" "${rec_model_value}")
+                            set_infer_params1=$(func_set_params "${infer_key1}" "${infer_value1}")
+                            command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_model_dir} ${set_batchsize} ${set_infer_params0} ${set_infer_data} ${set_benchmark} ${set_precision} ${set_infer_params1} > ${_save_log_path} 2>&1 "
+                            eval $command
+                            last_status=${PIPESTATUS[0]}
+                            eval "cat ${_save_log_path}"
+                            status_check $last_status "${command}" "${status_log}"
+                        done
+                    done
+                done
+            done
+        elif [ ${use_gpu} = "True" ] || [ ${use_gpu} = "gpu" ]; then
+            for use_trt in ${use_trt_list[*]}; do
+                for precision in ${precision_list[*]}; do
+                    if [[ ${_flag_quant} = "False" ]] && [[ ${precision} =~ "int8" ]]; then
+                        continue
+                    fi 
+                    if [[ ${precision} =~ "fp16" || ${precision} =~ "int8" ]] && [ ${use_trt} = "False" ]; then
+                        continue
+                    fi
+                    if [[ ${use_trt} = "False" || ${precision} =~ "int8" ]] && [ ${_flag_quant} = "True" ]; then
+                        continue
+                    fi
+                    for batch_size in ${batch_size_list[*]}; do
+                        _save_log_path="${_log_path}/python_infer_gpu_usetrt_${use_trt}_precision_${precision}_batchsize_${batch_size}.log"
+                        set_infer_data=$(func_set_params "${image_dir_key}" "${_img_dir}")
+                        set_benchmark=$(func_set_params "${benchmark_key}" "${benchmark_value}")
+                        set_batchsize=$(func_set_params "${batch_size_key}" "${batch_size}")
+                        set_tensorrt=$(func_set_params "${use_trt_key}" "${use_trt}")
+                        set_precision=$(func_set_params "${precision_key}" "${precision}")
+                        set_model_dir=$(func_set_params "${infer_model_key}" "${_model_dir}")
+                        set_infer_params0=$(func_set_params "${save_log_key}" "${save_log_value}")
+                        set_infer_params1=$(func_set_params "${infer_key1}" "${infer_value1}")
+                        command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${set_tensorrt} ${set_precision} ${set_model_dir} ${set_batchsize} ${set_infer_data} ${set_benchmark} ${set_infer_params1} ${set_infer_params0} > ${_save_log_path} 2>&1 "
+                        eval $command
+                        last_status=${PIPESTATUS[0]}
+                        eval "cat ${_save_log_path}"
+                        status_check $last_status "${command}" "${status_log}"
+                        
+                    done
+                done
+            done
+        else
+            echo "Does not support hardware other than CPU and GPU Currently!"
+        fi
+    done
+}
+
 if [ ${MODE} = "whole_infer" ]; then
     GPUID=$3
     if [ ${#GPUID} -le 0 ];then
@@ -68,7 +150,6 @@ if [ ${MODE} = "whole_infer" ]; then
             set_save_infer_key=$(func_set_params "${save_infer_key}" "${save_infer_dir}")
             export_cmd="${python} ${infer_run_exports[Count]} ${set_export_weight} ${set_save_infer_key}"
             echo ${infer_run_exports[Count]} 
-            echo  $export_cmd
             eval $export_cmd
             status_export=$?
             status_check $status_export "${export_cmd}" "${status_log}"
@@ -85,3 +166,4 @@ if [ ${MODE} = "whole_infer" ]; then
     done
 fi
 
+

test_tipc/test_lite_arm_cpp.sh

+#!/bin/bash
+source ./common_func.sh
+export LD_LIBRARY_PATH=${PWD}:$LD_LIBRARY_PATH
+
+FILENAME=$1
+dataline=$(cat $FILENAME)
+# parser params
+IFS=$'\n'
+lines=(${dataline})
+
+# parser lite inference
+inference_cmd=$(func_parser_value "${lines[1]}")
+runtime_device=$(func_parser_value "${lines[2]}")
+det_model_list=$(func_parser_value "${lines[3]}")
+rec_model_list=$(func_parser_value "${lines[4]}")
+cls_model_list=$(func_parser_value "${lines[5]}")
+cpu_threads_list=$(func_parser_value "${lines[6]}")
+det_batch_size_list=$(func_parser_value "${lines[7]}")
+rec_batch_size_list=$(func_parser_value "${lines[8]}")
+infer_img_dir_list=$(func_parser_value "${lines[9]}")
+config_dir=$(func_parser_value "${lines[10]}")
+rec_dict_dir=$(func_parser_value "${lines[11]}")
+benchmark_value=$(func_parser_value "${lines[12]}")
+
+if [[ $inference_cmd =~ "det" ]]; then
+    lite_model_list=${det_lite_model_list}
+elif [[ $inference_cmd =~ "rec" ]]; then
+    lite_model_list=(${rec_lite_model_list[*]} ${cls_lite_model_list[*]})
+elif [[ $inference_cmd =~ "system" ]]; then
+    lite_model_list=(${det_lite_model_list[*]} ${rec_lite_model_list[*]} ${cls_lite_model_list[*]})
+else
+    echo "inference_cmd is wrong, please check."
+    exit 1
+fi
+
+LOG_PATH="./output"
+mkdir -p ${LOG_PATH}
+status_log="${LOG_PATH}/results.log"
+
+
+function func_test_det(){
+    IFS='|'
+    _script=$1
+    _det_model=$2
+    _log_path=$3
+    _img_dir=$4
+    _config=$5
+    if [[ $_det_model =~ "slim" ]]; then
+        precision="INT8"
+    else
+        precision="FP32"
+    fi
+
+    # lite inference
+    for num_threads in ${cpu_threads_list[*]}; do
+	for det_batchsize in ${det_batch_size_list[*]}; do
+            _save_log_path="${_log_path}/lite_${_det_model}_runtime_device_${runtime_device}_precision_${precision}_det_batchsize_${det_batchsize}_threads_${num_threads}.log"
+            command="${_script} ${_det_model} ${runtime_device} ${precision} ${num_threads} ${det_batchsize}  ${_img_dir} ${_config} ${benchmark_value} > ${_save_log_path} 2>&1"
+            eval ${command}
+            status_check $? "${command}" "${status_log}"
+        done
+    done
+}
+
+function func_test_rec(){
+    IFS='|'
+    _script=$1
+    _rec_model=$2
+    _cls_model=$3
+    _log_path=$4
+    _img_dir=$5
+    _config=$6
+    _rec_dict_dir=$7
+
+    if [[ $_det_model =~ "slim" ]]; then
+        _precision="INT8"
+    else
+        _precision="FP32"
+    fi
+
+    # lite inference
+    for num_threads in ${cpu_threads_list[*]}; do
+	for rec_batchsize in ${rec_batch_size_list[*]}; do
+            _save_log_path="${_log_path}/lite_${_rec_model}_${cls_model}_runtime_device_${runtime_device}_precision_${_precision}_rec_batchsize_${rec_batchsize}_threads_${num_threads}.log"
+            command="${_script} ${_rec_model} ${_cls_model} ${runtime_device} ${_precision} ${num_threads} ${rec_batchsize}  ${_img_dir} ${_config} ${_rec_dict_dir} ${benchmark_value} > ${_save_log_path} 2>&1"
+            eval ${command}
+            status_check $? "${command}" "${status_log}"
+        done
+    done
+}
+
+function func_test_system(){
+    IFS='|'
+    _script=$1
+    _det_model=$2
+    _rec_model=$3
+    _cls_model=$4
+    _log_path=$5
+    _img_dir=$6
+    _config=$7
+    _rec_dict_dir=$8
+    if [[ $_det_model =~ "slim" ]]; then
+        _precision="INT8"
+    else
+        _precision="FP32"
+    fi
+
+    # lite inference
+    for num_threads in ${cpu_threads_list[*]}; do
+	for det_batchsize in ${det_batch_size_list[*]}; do
+	   for rec_batchsize in ${rec_batch_size_list[*]}; do
+                _save_log_path="${_log_path}/lite_${_det_model}_${_rec_model}_${_cls_model}_runtime_device_${runtime_device}_precision_${_precision}_det_batchsize_${det_batchsize}_rec_batchsize_${rec_batchsize}_threads_${num_threads}.log"
+                command="${_script} ${_det_model} ${_rec_model} ${_cls_model} ${runtime_device} ${_precision} ${num_threads} ${det_batchsize}  ${_img_dir} ${_config} ${_rec_dict_dir} ${benchmark_value} > ${_save_log_path} 2>&1"
+               eval ${command}
+               status_check $? "${command}" "${status_log}"
+	    done
+        done
+    done
+}
+
+
+echo "################### run test ###################"
+
+if [[ $inference_cmd =~ "det" ]]; then
+    IFS="|"
+    det_model_list=(${det_model_list[*]})
+
+    for i in {0..1}; do
+        #run lite inference
+        for img_dir in ${infer_img_dir_list[*]}; do
+            func_test_det "${inference_cmd}" "${det_model_list[i]}_opt.nb" "${LOG_PATH}" "${img_dir}" "${config_dir}"
+        done
+    done
+
+elif [[ $inference_cmd =~ "rec" ]]; then
+    IFS="|"
+    rec_model_list=(${rec_model_list[*]})
+    cls_model_list=(${cls_model_list[*]})
+
+    for i in {0..1}; do
+        #run lite inference
+        for img_dir in ${infer_img_dir_list[*]}; do
+            func_test_rec "${inference_cmd}" "${rec_model}_opt.nb" "${cls_model_list[i]}_opt.nb" "${LOG_PATH}" "${img_dir}" "${rec_dict_dir}" "${config_dir}"
+        done
+    done
+
+elif [[ $inference_cmd =~ "system" ]]; then
+    IFS="|"
+    det_model_list=(${det_model_list[*]})
+    rec_model_list=(${rec_model_list[*]})
+    cls_model_list=(${cls_model_list[*]})
+
+    for i in {0..1}; do
+	#run lite inference
+        for img_dir in ${infer_img_dir_list[*]}; do
+            func_test_system "${inference_cmd}" "${det_model_list[i]}_opt.nb" "${rec_model_list[i]}_opt.nb" "${cls_model_list[i]}_opt.nb" "${LOG_PATH}" "${img_dir}" "${config_dir}" "${rec_dict_dir}"
+        done
+    done
+fi

test_tipc/test_lite_arm_cpu_cpp.sh

-#!/bin/bash
-source ./common_func.sh
-export LD_LIBRARY_PATH=${PWD}:$LD_LIBRARY_PATH
-
-FILENAME=$1
-dataline=$(cat $FILENAME)
-# parser params
-IFS=$'\n'
-lines=(${dataline})
-
-# parser lite inference
-lite_inference_cmd=$(func_parser_value "${lines[1]}")
-lite_model_dir_list=$(func_parser_value "${lines[2]}")
-runtime_device=$(func_parser_value "${lines[3]}")
-lite_cpu_threads_list=$(func_parser_value "${lines[4]}")
-lite_batch_size_list=$(func_parser_value "${lines[5]}")
-lite_infer_img_dir_list=$(func_parser_value "${lines[8]}")
-lite_config_dir=$(func_parser_value "${lines[9]}")
-lite_rec_dict_dir=$(func_parser_value "${lines[10]}")
-lite_benchmark_value=$(func_parser_value "${lines[11]}")
-
-
-LOG_PATH="./output"
-mkdir -p ${LOG_PATH}
-status_log="${LOG_PATH}/results.log"
-
-
-function func_lite(){
-    IFS='|'
-    _script=$1
-    _lite_model=$2
-    _log_path=$3
-    _img_dir=$4
-    _config=$5
-    if [[ $lite_model =~ "slim" ]]; then
-        precision="INT8"
-    else
-        precision="FP32"
-    fi
-
-    # lite inference
-    for num_threads in ${lite_cpu_threads_list[*]}; do
-	for batchsize in ${lite_batch_size_list[*]}; do
-            _save_log_path="${_log_path}/lite_${_lite_model}_runtime_device_${runtime_device}_precision_${precision}_batchsize_${batchsize}_threads_${num_threads}.log"
-            command="${_script} ${_lite_model} ${runtime_device} ${precision} ${num_threads} ${batchsize}  ${_img_dir} ${_config} ${lite_benchmark_value} > ${_save_log_path} 2>&1"
-            eval ${command}
-            status_check $? "${command}" "${status_log}"
-        done
-    done
-}
-
-
-echo "################### run test ###################"
-IFS="|"
-for lite_model in ${lite_model_dir_list[*]}; do
-    #run lite inference
-    for img_dir in ${lite_infer_img_dir_list[*]}; do
-        func_lite "${lite_inference_cmd}" "${lite_model}_opt.nb" "${LOG_PATH}" "${img_dir}" "${lite_config_dir}"
-    done
-done

test_tipc/test_serving.sh

@@ -10,7 +10,7 @@ lines=(${dataline})
 
 # parser serving
 model_name=$(func_parser_value "${lines[1]}")
-python=$(func_parser_value "${lines[2]}")
+python_list=$(func_parser_value "${lines[2]}")
 trans_model_py=$(func_parser_value "${lines[3]}")
 infer_model_dir_key=$(func_parser_key "${lines[4]}")
 infer_model_dir_value=$(func_parser_value "${lines[4]}")
@@ -54,14 +54,15 @@ function func_serving(){
     set_serving_server=$(func_set_params "${serving_server_key}" "${serving_server_value}")
     set_serving_client=$(func_set_params "${serving_client_key}" "${serving_client_value}")
     set_image_dir=$(func_set_params "${image_dir_key}" "${image_dir_value}")
-    trans_model_cmd="${python} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client}"
+    python_list=(${python_list})
+    trans_model_cmd="${python_list[0]} ${trans_model_py} ${set_dirname} ${set_model_filename} ${set_params_filename} ${set_serving_server} ${set_serving_client}"
     eval $trans_model_cmd
     cd ${serving_dir_value}
     echo $PWD
     unset https_proxy
     unset http_proxy
-    for python in ${python[*]}; do
-        if [ ${python} = "cpp"]; then
+    for python in ${python_list[*]}; do
+        if [ ${python} = "cpp" ]; then
             for use_gpu in ${web_use_gpu_list[*]}; do
                 if [ ${use_gpu} = "null" ]; then
                     web_service_cpp_cmd="${python} -m paddle_serving_server.serve --model ppocr_det_mobile_2.0_serving/ ppocr_rec_mobile_2.0_serving/ --port 9293"
@@ -91,9 +92,6 @@ function func_serving(){
                 echo ${ues_gpu}
                 if [ ${use_gpu} = "null" ]; then
                     for use_mkldnn in ${web_use_mkldnn_list[*]}; do
-                        if [ ${use_mkldnn} = "False" ]; then
-                            continue
-                        fi
                         for threads in ${web_cpu_threads_list[*]}; do
                             set_cpu_threads=$(func_set_params "${web_cpu_threads_key}" "${threads}")
                             web_service_cmd="${python} ${web_service_py} ${web_use_gpu_key}=${use_gpu} ${web_use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} &"
@@ -124,6 +122,9 @@ function func_serving(){
                                 continue
                             fi
                             set_tensorrt=$(func_set_params "${web_use_trt_key}" "${use_trt}")
+                            if [ ${use_trt} = True ]; then
+                                device_type=2
+                            fi
                             set_precision=$(func_set_params "${web_precision_key}" "${precision}")
                             web_service_cmd="${python} ${web_service_py} ${web_use_gpu_key}=${use_gpu} ${set_tensorrt} ${set_precision} & "
                             eval $web_service_cmd

test_tipc/test_train_inference_python.sh

@@ -20,10 +20,10 @@ train_use_gpu_value=$(func_parser_value "${lines[4]}")
 autocast_list=$(func_parser_value "${lines[5]}")
 autocast_key=$(func_parser_key "${lines[5]}")
 epoch_key=$(func_parser_key "${lines[6]}")
-epoch_num=$(func_parser_params "${lines[6]}")
+epoch_num=$(func_parser_params "${lines[6]}" "${MODE}")
 save_model_key=$(func_parser_key "${lines[7]}")
 train_batch_key=$(func_parser_key "${lines[8]}")
-train_batch_value=$(func_parser_params "${lines[8]}")
+train_batch_value=$(func_parser_params "${lines[8]}" "${MODE}")
 pretrain_model_key=$(func_parser_key "${lines[9]}")
 pretrain_model_value=$(func_parser_value "${lines[9]}")
 train_model_name=$(func_parser_value "${lines[10]}")
@@ -90,34 +90,39 @@ infer_value1=$(func_parser_value "${lines[50]}")
 
 # parser klquant_infer
 if [ ${MODE} = "klquant_whole_infer" ]; then
-    dataline=$(awk 'NR==85 NR==101{print}'  $FILENAME)
+    dataline=$(awk 'NR==1, NR==17{print}'  $FILENAME)
     lines=(${dataline})
+    model_name=$(func_parser_value "${lines[1]}")
+    python=$(func_parser_value "${lines[2]}")
+    export_weight=$(func_parser_key "${lines[3]}")
+    save_infer_key=$(func_parser_key "${lines[4]}")
     # parser inference model 
-    infer_model_dir_list=$(func_parser_value "${lines[1]}")
-    infer_export_list=$(func_parser_value "${lines[2]}")
-    infer_is_quant=$(func_parser_value "${lines[3]}")
+    infer_model_dir_list=$(func_parser_value "${lines[5]}")
+    infer_export_list=$(func_parser_value "${lines[6]}")
+    infer_is_quant=$(func_parser_value "${lines[7]}")
     # parser inference 
-    inference_py=$(func_parser_value "${lines[4]}")
-    use_gpu_key=$(func_parser_key "${lines[5]}")
-    use_gpu_list=$(func_parser_value "${lines[5]}")
-    use_mkldnn_key=$(func_parser_key "${lines[6]}")
-    use_mkldnn_list=$(func_parser_value "${lines[6]}")
-    cpu_threads_key=$(func_parser_key "${lines[7]}")
-    cpu_threads_list=$(func_parser_value "${lines[7]}")
-    batch_size_key=$(func_parser_key "${lines[8]}")
-    batch_size_list=$(func_parser_value "${lines[8]}")
-    use_trt_key=$(func_parser_key "${lines[9]}")
-    use_trt_list=$(func_parser_value "${lines[9]}")
-    precision_key=$(func_parser_key "${lines[10]}")
-    precision_list=$(func_parser_value "${lines[10]}")
-    infer_model_key=$(func_parser_key "${lines[11]}")
-    image_dir_key=$(func_parser_key "${lines[12]}")
-    infer_img_dir=$(func_parser_value "${lines[12]}")
-    save_log_key=$(func_parser_key "${lines[13]}")
-    benchmark_key=$(func_parser_key "${lines[14]}")
-    benchmark_value=$(func_parser_value "${lines[14]}")
-    infer_key1=$(func_parser_key "${lines[15]}")
-    infer_value1=$(func_parser_value "${lines[15]}")
+    inference_py=$(func_parser_value "${lines[8]}")
+    use_gpu_key=$(func_parser_key "${lines[9]}")
+    use_gpu_list=$(func_parser_value "${lines[9]}")
+    use_mkldnn_key=$(func_parser_key "${lines[10]}")
+    use_mkldnn_list=$(func_parser_value "${lines[10]}")
+    cpu_threads_key=$(func_parser_key "${lines[11]}")
+    cpu_threads_list=$(func_parser_value "${lines[11]}")
+    batch_size_key=$(func_parser_key "${lines[12]}")
+    batch_size_list=$(func_parser_value "${lines[12]}")
+    use_trt_key=$(func_parser_key "${lines[13]}")
+    use_trt_list=$(func_parser_value "${lines[13]}")
+    precision_key=$(func_parser_key "${lines[14]}")
+    precision_list=$(func_parser_value "${lines[14]}")
+    infer_model_key=$(func_parser_key "${lines[15]}")
+    image_dir_key=$(func_parser_key "${lines[16]}")
+    infer_img_dir=$(func_parser_value "${lines[16]}")
+    save_log_key=$(func_parser_key "${lines[17]}")
+    save_log_value=$(func_parser_value "${lines[17]}")
+    benchmark_key=$(func_parser_key "${lines[18]}")
+    benchmark_value=$(func_parser_value "${lines[18]}")
+    infer_key1=$(func_parser_key "${lines[19]}")
+    infer_value1=$(func_parser_value "${lines[19]}")
 fi
 
 LOG_PATH="./test_tipc/output"
@@ -157,8 +162,9 @@ function func_inference(){
                             set_batchsize=$(func_set_params "${batch_size_key}" "${batch_size}")
                             set_cpu_threads=$(func_set_params "${cpu_threads_key}" "${threads}")
                             set_model_dir=$(func_set_params "${infer_model_key}" "${_model_dir}")
+                            set_infer_params0=$(func_set_params "${save_log_key}" "${save_log_value}")
                             set_infer_params1=$(func_set_params "${infer_key1}" "${infer_value1}")
-                            command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_model_dir} ${set_batchsize} ${set_infer_data} ${set_benchmark} ${set_precision} ${set_infer_params1} > ${_save_log_path} 2>&1 "
+                            command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${use_mkldnn_key}=${use_mkldnn} ${set_cpu_threads} ${set_model_dir} ${set_batchsize} ${set_infer_params0} ${set_infer_data} ${set_benchmark} ${set_precision} ${set_infer_params1} > ${_save_log_path} 2>&1 "
                             eval $command
                             last_status=${PIPESTATUS[0]}
                             eval "cat ${_save_log_path}"
@@ -187,8 +193,9 @@ function func_inference(){
                         set_tensorrt=$(func_set_params "${use_trt_key}" "${use_trt}")
                         set_precision=$(func_set_params "${precision_key}" "${precision}")
                         set_model_dir=$(func_set_params "${infer_model_key}" "${_model_dir}")
+                        set_infer_params0=$(func_set_params "${save_log_key}" "${save_log_value}")
                         set_infer_params1=$(func_set_params "${infer_key1}" "${infer_value1}")
-                        command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${set_tensorrt} ${set_precision} ${set_model_dir} ${set_batchsize} ${set_infer_data} ${set_benchmark} ${set_infer_params1} > ${_save_log_path} 2>&1 "
+                        command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${set_tensorrt} ${set_precision} ${set_model_dir} ${set_batchsize} ${set_infer_data} ${set_benchmark} ${set_infer_params1} ${set_infer_params0} > ${_save_log_path} 2>&1 "
                         eval $command
                         last_status=${PIPESTATUS[0]}
                         eval "cat ${_save_log_path}"
@@ -224,7 +231,7 @@ if [ ${MODE} = "whole_infer" ] || [ ${MODE} = "klquant_whole_infer" ]; then
             set_save_infer_key=$(func_set_params "${save_infer_key}" "${save_infer_dir}")
             export_cmd="${python} ${infer_run_exports[Count]} ${set_export_weight} ${set_save_infer_key}"
             echo ${infer_run_exports[Count]} 
-            echo  $export_cmd
+            echo $export_cmd
             eval $export_cmd
             status_export=$?
             status_check $status_export "${export_cmd}" "${status_log}"
@@ -233,7 +240,7 @@ if [ ${MODE} = "whole_infer" ] || [ ${MODE} = "klquant_whole_infer" ]; then
         fi
         #run inference
         is_quant=${infer_quant_flag[Count]}
-        if [ ${MODE} = "klquant_infer" ]; then
+        if [ ${MODE} = "klquant_whole_infer" ]; then
             is_quant="True"
         fi
         func_inference "${python}" "${inference_py}" "${save_infer_dir}" "${LOG_PATH}" "${infer_img_dir}" ${is_quant}
@@ -334,7 +341,7 @@ else
 
                 set_eval_pretrain=$(func_set_params "${pretrain_model_key}" "${save_log}/${train_model_name}")
                 # save norm trained models to set pretrain for pact training and fpgm training 
-                if [ ${trainer} = ${trainer_norm} ] && [ ${nodes} -le 1]; then
+                if [ ${trainer} = ${trainer_norm} ] && [ ${nodes} -le 1 ]; then
                     load_norm_train_model=${set_eval_pretrain}
                 fi
                 # run eval 
@@ -357,7 +364,7 @@ else
                     #run inference
                     eval $env
                     save_infer_path="${save_log}"
-                    if [ ${inference_dir} != "null" ] && [ ${inference_dir} != '##' ]; then
+                    if [[ ${inference_dir} != "null" ]] && [[ ${inference_dir} != '##' ]]; then
                         infer_model_dir="${save_infer_path}/${inference_dir}"
                     else
                         infer_model_dir=${save_infer_path}

tools/infer/predict_e2e.py

@@ -68,7 +68,6 @@ class TextE2E(object):
             postprocess_params["character_dict_path"] = args.e2e_char_dict_path
             postprocess_params["valid_set"] = args.e2e_pgnet_valid_set
             postprocess_params["mode"] = args.e2e_pgnet_mode
-            self.e2e_pgnet_polygon = args.e2e_pgnet_polygon
         else:
             logger.info("unknown e2e_algorithm:{}".format(self.e2e_algorithm))
             sys.exit(0)

tools/infer/predict_rec.py

@@ -91,7 +91,7 @@ class TextRecognizer(object):
                 time_keys=[
                     'preprocess_time', 'inference_time', 'postprocess_time'
                 ],
-                warmup=2,
+                warmup=0,
                 logger=logger)
 
     def resize_norm_img(self, img, max_wh_ratio):

tools/infer/predict_system.py

@@ -49,11 +49,19 @@ class TextSystem(object):
         if self.use_angle_cls:
             self.text_classifier = predict_cls.TextClassifier(args)
 
-    def print_draw_crop_rec_res(self, img_crop_list, rec_res):
+        self.args = args
+        self.crop_image_res_index = 0
+
+    def draw_crop_rec_res(self, output_dir, img_crop_list, rec_res):
+        os.makedirs(output_dir, exist_ok=True)
         bbox_num = len(img_crop_list)
         for bno in range(bbox_num):
-            cv2.imwrite("./output/img_crop_%d.jpg" % bno, img_crop_list[bno])
-            logger.info(bno, rec_res[bno])
+            cv2.imwrite(
+                os.path.join(output_dir,
+                             f"mg_crop_{bno+self.crop_image_res_index}.jpg"),
+                img_crop_list[bno])
+            logger.debug(f"{bno}, {rec_res[bno]}")
+        self.crop_image_res_index += bbox_num
 
     def __call__(self, img, cls=True):
         ori_im = img.copy()
@@ -80,7 +88,9 @@ class TextSystem(object):
         rec_res, elapse = self.text_recognizer(img_crop_list)
         logger.debug("rec_res num  : {}, elapse : {}".format(
             len(rec_res), elapse))
-        # self.print_draw_crop_rec_res(img_crop_list, rec_res)
+        if self.args.save_crop_res:
+            self.draw_crop_rec_res(self.args.crop_res_save_dir, img_crop_list,
+                                   rec_res)
         filter_boxes, filter_rec_res = [], []
         for box, rec_reuslt in zip(dt_boxes, rec_res):
             text, score = rec_reuslt
@@ -135,17 +145,17 @@ def main(args):
         if not flag:
             img = cv2.imread(image_file)
         if img is None:
-            logger.info("error in loading image:{}".format(image_file))
+            logger.debug("error in loading image:{}".format(image_file))
             continue
         starttime = time.time()
         dt_boxes, rec_res = text_sys(img)
         elapse = time.time() - starttime
         total_time += elapse
 
-        logger.info(
+        logger.debug(
             str(idx) + "  Predict time of %s: %.3fs" % (image_file, elapse))
         for text, score in rec_res:
-            logger.info("{}, {:.3f}".format(text, score))
+            logger.debug("{}, {:.3f}".format(text, score))
 
         if is_visualize:
             image = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
@@ -160,19 +170,17 @@ def main(args):
                 scores,
                 drop_score=drop_score,
                 font_path=font_path)
-            draw_img_save = "./inference_results/"
-            if not os.path.exists(draw_img_save):
-                os.makedirs(draw_img_save)
+            draw_img_save_dir = args.draw_img_save_dir
+            os.makedirs(draw_img_save_dir, exist_ok=True)
             if flag:
                 image_file = image_file[:-3] + "png"
             cv2.imwrite(
-                os.path.join(draw_img_save, os.path.basename(image_file)),
+                os.path.join(draw_img_save_dir, os.path.basename(image_file)),
                 draw_img[:, :, ::-1])
-            logger.info("The visualized image saved in {}".format(
-                os.path.join(draw_img_save, os.path.basename(image_file))))
+            logger.debug("The visualized image saved in {}".format(
+                os.path.join(draw_img_save_dir, os.path.basename(image_file))))
 
     logger.info("The predict total time is {}".format(time.time() - _st))
-    logger.info("\nThe predict total time is {}".format(total_time))
     if args.benchmark:
         text_sys.text_detector.autolog.report()
         text_sys.text_recognizer.autolog.report()

tools/infer/utility.py

@@ -96,7 +96,6 @@ def init_args():
     parser.add_argument(
         "--e2e_char_dict_path", type=str, default="./ppocr/utils/ic15_dict.txt")
     parser.add_argument("--e2e_pgnet_valid_set", type=str, default='totaltext')
-    parser.add_argument("--e2e_pgnet_polygon", type=str2bool, default=True)
     parser.add_argument("--e2e_pgnet_mode", type=str, default='fast')
 
     # params for text classifier
@@ -110,7 +109,13 @@ def init_args():
     parser.add_argument("--enable_mkldnn", type=str2bool, default=False)
     parser.add_argument("--cpu_threads", type=int, default=10)
     parser.add_argument("--use_pdserving", type=str2bool, default=False)
-    parser.add_argument("--warmup", type=str2bool, default=True)
+    parser.add_argument("--warmup", type=str2bool, default=False)
+
+    #
+    parser.add_argument(
+        "--draw_img_save_dir", type=str, default="./inference_results")
+    parser.add_argument("--save_crop_res", type=str2bool, default=False)
+    parser.add_argument("--crop_res_save_dir", type=str, default="./output")
 
     # multi-process
     parser.add_argument("--use_mp", type=str2bool, default=False)
@@ -185,6 +190,7 @@ def create_predictor(args, mode, logger):
             config.enable_use_gpu(args.gpu_mem, 0)
             if args.use_tensorrt:
                 config.enable_tensorrt_engine(
+                    workspace_size=1 << 30,
                     precision_mode=precision,
                     max_batch_size=args.max_batch_size,
                     min_subgraph_size=args.min_subgraph_size)
@@ -205,7 +211,7 @@ def create_predictor(args, mode, logger):
                     "nearest_interp_v2_0.tmp_0": [1, 256, 2, 2]
                 }
                 max_input_shape = {
-                    "x": [1, 3, 1280, 1280],
+                    "x": [1, 3, 1536, 1536],
                     "conv2d_92.tmp_0": [1, 120, 400, 400],
                     "conv2d_91.tmp_0": [1, 24, 200, 200],
                     "conv2d_59.tmp_0": [1, 96, 400, 400],
@@ -255,7 +261,7 @@ def create_predictor(args, mode, logger):
                 opt_input_shape.update(opt_pact_shape)
             elif mode == "rec":
                 min_input_shape = {"x": [1, 3, 32, 10]}
-                max_input_shape = {"x": [args.rec_batch_num, 3, 32, 1024]}
+                max_input_shape = {"x": [args.rec_batch_num, 3, 32, 1536]}
                 opt_input_shape = {"x": [args.rec_batch_num, 3, 32, 320]}
             elif mode == "cls":
                 min_input_shape = {"x": [1, 3, 48, 10]}
@@ -305,11 +311,10 @@ def create_predictor(args, mode, logger):
 
 
 def get_infer_gpuid():
-    cmd = "nvidia-smi"
-    res = os.popen(cmd).readlines()
-    if len(res) == 0:
-        return None
-    cmd = "env | grep CUDA_VISIBLE_DEVICES"
+    if not paddle.fluid.core.is_compiled_with_rocm():
+        cmd = "env | grep CUDA_VISIBLE_DEVICES"
+    else:
+        cmd = "env | grep HIP_VISIBLE_DEVICES"
     env_cuda = os.popen(cmd).readlines()
     if len(env_cuda) == 0:
         return 0

tools/infer_det.py

@@ -53,6 +53,7 @@ def draw_det_res(dt_boxes, config, img, img_name, save_path):
         logger.info("The detected Image saved in {}".format(save_path))
 
 
+@paddle.no_grad()
 def main():
     global_config = config['Global']