Skip to content

M
models

PaddlePaddle / models
大约 1 年 前同步成功

通知 222
Star 6828
Fork 2962
M
models
未验证 提交 ef7ebd08 编写于 作者: C cuicheng01 提交者： GitHub
浏览文件
操作

update tipc lite docs (#5390) 

* update tipc lite docs

* update tipc lite docs
上级 f5c8dac5
隐藏空白更改
内联 并排
Showing with 152 addition and 53 deletion
docs/tipc_test/development_specification_docs/Lite_arm_cpu_cpp_infer.md
浏览文件 @ ef7ebd08
...@@ -56,14 +56,14 @@ Lite预测测试工具位于PaddleOCR dygraph分支下的test_tipc目录，与Li ...@@ -56,14 +56,14 @@ Lite预测测试工具位于PaddleOCR dygraph分支下的test_tipc目录，与Li
test_tipc/ test_tipc/
├── common_func.sh ├── common_func.sh
├── configs # 配置文件目录 ├── configs # 配置文件目录
│   ├── ppocr_det_mobile │   ├── ch_PP-OCRv2_det
│   │   ├── model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt │   │   ├── model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt
│   │   ├── ... │   │   ├── ...
│ ├── ... │ ├── ...
│   ├── ppocr_system_mobile │   ├── ch_PP-OCRv2
│   │   ├── model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt │   │   ├── model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt
│   │   └── ... │   │   └── ...
├── prepare_lite_cpp.sh # 完成test_cpu_cpp.sh运行所需要的数据和模型下载 ├── prepare_lite_cpp.sh # 完成test_lite_**.sh运行所需要的数据和模型下载
├── test_lite_arm_cpp.sh # lite测试主程序 ├── test_lite_arm_cpp.sh # lite测试主程序
``` ```
...@@ -89,9 +89,18 @@ Lite端ARM_CPU预测接入TIPC包含如下三个步骤，接下来将依次介 ...@@ -89,9 +89,18 @@ Lite端ARM_CPU预测接入TIPC包含如下三个步骤，接下来将依次介
以PaddleOCR文本检测模型为例，使用方式： 以PaddleOCR文本检测模型为例，使用方式：
下载Paddle-Lite预测库的方式：
```
bash test_tipc/prepare_lite_arm_cpp.sh test_tipc/configs/ch_PP-OCRv2_det/model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt download
```
编译Paddle-Lite预测库的方式：
``` ```
bash test_tipc/prepare_lite_arm_cpp.sh test_tipc/configs/ppocr_det_mobile/model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt bash test_tipc/prepare_lite_arm_cpp.sh test_tipc/configs/ch_PP-OCRv2_det/model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt compile
``` ```
`prepare_lite_arm_cpp.sh`具体内容： `prepare_lite_arm_cpp.sh`具体内容：
1.解析`model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt`部分用于预测的参数字段，方便后续预测。 1.解析`model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt`部分用于预测的参数字段，方便后续预测。
...@@ -104,6 +113,7 @@ dataline=$(cat ${FILENAME}) ...@@ -104,6 +113,7 @@ dataline=$(cat ${FILENAME})
IFS=$'\n' IFS=$'\n'
lines=(${dataline}) lines=(${dataline})
IFS=$'\n' IFS=$'\n'
paddlelite_library_source=$2
inference_cmd=$(func_parser_value "${lines[1]}") inference_cmd=$(func_parser_value "${lines[1]}")
DEVICE=$(func_parser_value "${lines[2]}") DEVICE=$(func_parser_value "${lines[2]}")
...@@ -112,10 +122,29 @@ rec_lite_model_list=$(func_parser_value "${lines[4]}") ...@@ -112,10 +122,29 @@ rec_lite_model_list=$(func_parser_value "${lines[4]}")
cls_lite_model_list=$(func_parser_value "${lines[5]}") cls_lite_model_list=$(func_parser_value "${lines[5]}")
``` ```
2.转换`infernce model`到Lite预测的`.nb`模型 2.填写不同硬件设备需要的信息
```
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
```
3.转换`infernce model`到Lite预测的`.nb`模型
``` ```
# prepare lite .nb model # prepare paddlelite model
if [[ $inference_cmd =~ "det" ]];then if [[ $inference_cmd =~ "det" ]];then
lite_model_list=${det_lite_model_list} lite_model_list=${det_lite_model_list}
elif [[ $inference_cmd =~ "rec" ]];then elif [[ $inference_cmd =~ "rec" ]];then
...@@ -127,6 +156,11 @@ else ...@@ -127,6 +156,11 @@ else
exit 1 exit 1
fi fi
pip install paddlelite==2.10-rc
current_dir=${PWD}
IFS="|"
model_path=./inference_models
for model in ${lite_model_list[*]}; do for model in ${lite_model_list[*]}; do
if [[ $model =~ "PP-OCRv2" ]];then if [[ $model =~ "PP-OCRv2" ]];then
inference_model_url=https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/${model}.tar inference_model_url=https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/${model}.tar
...@@ -146,37 +180,51 @@ for model in ${lite_model_list[*]}; do ...@@ -146,37 +180,51 @@ for model in ${lite_model_list[*]}; do
done done
``` ```
3.准备测试数据 4.准备测试数据
``` ```
data_url=https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/icdar2015_lite.tar data_url=https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/icdar2015_lite.tar
model_path=./inference_models
inference_model=${inference_model_url##*/}
data_file=${data_url##*/} data_file=${data_url##*/}
wget -nc -P ./inference_models ${inference_model_url}
wget -nc -P ./test_data ${data_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 ../ cd ./test_data && tar -xf ${data_file} && rm ${data_file} && cd ../
``` ```
4.准备Lite预测环境，此处需要下载或者编译Paddle-Lite预测库。 5.准备Lite预测环境，此处需要下载或者编译Paddle-Lite预测库。
``` ```
# prepare lite env # prepare lite env
paddlelite_zipfile=$(echo $paddlelite_url | awk -F "/" '{print $NF}') if [[ ${paddlelite_library_source} = "download" ]]; then
paddlelite_file=${paddlelite_zipfile:0:${end_index}} paddlelite_library_zipfile=$(echo $paddlelite_library_url | awk -F "/" '{print $NF}')
wget ${paddlelite_url} && tar -xf ${paddlelite_zipfile} paddlelite_library_file=${paddlelite_library_zipfile:0:${end_index}}
mkdir -p ${paddlelite_file}/demo/cxx/ocr/test_lite wget ${paddlelite_library_url} && tar -xf ${paddlelite_library_zipfile}
cp -r ${model_path}/*_opt.nb test_data ${paddlelite_file}/demo/cxx/ocr/test_lite cd ${paddlelite_library_zipfile}
cp ppocr/utils/ppocr_keys_v1.txt deploy/lite/config.txt ${paddlelite_file}/demo/cxx/ocr/test_lite elif [[ ${paddlelite_library_source} = "compile" ]]; then
cp -r ./deploy/lite/* ${paddlelite_file}/demo/cxx/ocr/ git clone -b release/v2.10 https://github.com/PaddlePaddle/Paddle-Lite.git
cp ${paddlelite_file}/cxx/lib/libpaddle_light_api_shared.so ${paddlelite_file}/demo/cxx/ocr/test_lite cd Paddle-Lite
cp ${FILENAME} test_tipc/test_lite_arm_cpp.sh test_tipc/common_func.sh ${paddlelite_file}/demo/cxx/ocr/test_lite ./lite/tools/build_android.sh --arch=armv8 --with_cv=ON --with_extra=ON --toolchain=clang --with_opencl=${compile_with_opencl}
cd ${paddlelite_file}/demo/cxx/ocr/ 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
```
6.准备编译以及后续在Lite设备上测试需要的文件
```
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 git clone https://github.com/cuicheng01/AutoLog.git
``` ```
5.交叉编译获得在手机上的可以运行的可执行文件
7.交叉编译获得在手机上的可以运行的可执行文件及删除当前目录下无用文件
``` ```
make -j make -j
...@@ -184,9 +232,11 @@ sleep 1 ...@@ -184,9 +232,11 @@ sleep 1
make -j make -j
cp ocr_db_crnn test_lite && cp test_lite/libpaddle_light_api_shared.so test_lite/libc++_shared.so 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} tar -cf test_lite.tar ./test_lite && cp test_lite.tar ${current_dir} && cd ${current_dir}
rm -rf ${paddlelite_file}* && rm -rf ${model_path} rm -rf ${paddlelite_library_file}* && rm -rf ${model_path}
``` ```
运行结果会在当前目录上生成test_lite.tar，里边的内容大致如下： 运行结果会在当前目录上生成test_lite.tar，里边的内容大致如下：
``` ```
├── common_func.sh # 通用函数，如解析参数等 ├── common_func.sh # 通用函数，如解析参数等
......
docs/tipc_test/development_specification_docs/Lite_arm_gpu_opencl_cpp_infer.md
浏览文件 @ ef7ebd08
...@@ -56,14 +56,14 @@ Lite预测测试工具位于PaddleOCR dygraph分支下的test_tipc目录，与Li ...@@ -56,14 +56,14 @@ Lite预测测试工具位于PaddleOCR dygraph分支下的test_tipc目录，与Li
test_tipc/ test_tipc/
├── common_func.sh ├── common_func.sh
├── configs # 配置文件目录 ├── configs # 配置文件目录
│   ├── ppocr_det_mobile │   ├── ch_PP-OCRv2_det
│   │   ├── model_linux_gpu_normal_normal_lite_cpp_arm_gpu_opencl.txt │   │   ├── model_linux_gpu_normal_normal_lite_cpp_arm_gpu_opencl.txt
│   │   ├── ... │   │   ├── ...
│ ├── ... │ ├── ...
│   ├── ppocr_system_mobile │   ├── ch_PP-OCRv2
│   │   ├── model_linux_gpu_normal_normal_lite_cpp_arm_gpu_opencl.txt │   │   ├── model_linux_gpu_normal_normal_lite_cpp_arm_gpu_opencl.txt
│   │   └── ... │   │   └── ...
├── prepare_lite_cpp.sh # 完成test_arm_cpp.sh运行所需要的数据和模型下载 ├── prepare_lite_cpp.sh # 完成test_lite_**.sh运行所需要的数据和模型下载
├── test_lite_arm_cpp.sh # lite测试主程序 ├── test_lite_arm_cpp.sh # lite测试主程序
``` ```
...@@ -89,12 +89,22 @@ Lite端ARM_GPU_OPENCL预测接入TIPC包含如下三个步骤，接下来将依 ...@@ -89,12 +89,22 @@ Lite端ARM_GPU_OPENCL预测接入TIPC包含如下三个步骤，接下来将依
以PaddleOCR文本检测模型为例，使用方式： 以PaddleOCR文本检测模型为例，使用方式：
下载Paddle-Lite预测库的方式：
``` ```
bash test_tipc/prepare_lite_arm_cpp.sh test_tipc/configs/ppocr_det_mobile/model_linux_gpu_normal_normal_lite_cpp_arm_gpu_opencl.txt bash test_tipc/prepare_lite_arm_cpp.sh test_tipc/configs/ch_PP-OCRv2_det/model_linux_gpu_normal_normal_lite_cpp_arm_gpu_opencl.txt download
``` ```
编译Paddle-Lite预测库的方式：
```
bash test_tipc/prepare_lite_arm_cpp.sh test_tipc/configs/ch_PP-OCRv2_det/model_linux_gpu_normal_normal_lite_cpp_arm_gpu_opencl.txt compile
```
`prepare_lite_arm_cpp.sh`具体内容： `prepare_lite_arm_cpp.sh`具体内容：
1.解析`model_linux_gpu_normal_normal_lite_cpp_arm_gpu_opencl.txt`部分用于预测的参数字段，方便后续预测。 1.解析`model_linux_gpu_normal_normal_lite_cpp_arm_cpu.txt`部分用于预测的参数字段，方便后续预测。
``` ```
source ./test_tipc/common_func.sh source ./test_tipc/common_func.sh
...@@ -104,6 +114,7 @@ dataline=$(cat ${FILENAME}) ...@@ -104,6 +114,7 @@ dataline=$(cat ${FILENAME})
IFS=$'\n' IFS=$'\n'
lines=(${dataline}) lines=(${dataline})
IFS=$'\n' IFS=$'\n'
paddlelite_library_source=$2
inference_cmd=$(func_parser_value "${lines[1]}") inference_cmd=$(func_parser_value "${lines[1]}")
DEVICE=$(func_parser_value "${lines[2]}") DEVICE=$(func_parser_value "${lines[2]}")
...@@ -112,10 +123,29 @@ rec_lite_model_list=$(func_parser_value "${lines[4]}") ...@@ -112,10 +123,29 @@ rec_lite_model_list=$(func_parser_value "${lines[4]}")
cls_lite_model_list=$(func_parser_value "${lines[5]}") cls_lite_model_list=$(func_parser_value "${lines[5]}")
``` ```
2.转换`infernce model`到Lite预测的`.nb`模型 2.填写不同硬件设备需要的信息
```
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
```
3.转换`infernce model`到Lite预测的`.nb`模型
``` ```
# prepare lite .nb model # prepare paddlelite model
if [[ $inference_cmd =~ "det" ]];then if [[ $inference_cmd =~ "det" ]];then
lite_model_list=${det_lite_model_list} lite_model_list=${det_lite_model_list}
elif [[ $inference_cmd =~ "rec" ]];then elif [[ $inference_cmd =~ "rec" ]];then
...@@ -127,6 +157,11 @@ else ...@@ -127,6 +157,11 @@ else
exit 1 exit 1
fi fi
pip install paddlelite==2.10-rc
current_dir=${PWD}
IFS="|"
model_path=./inference_models
for model in ${lite_model_list[*]}; do for model in ${lite_model_list[*]}; do
if [[ $model =~ "PP-OCRv2" ]];then if [[ $model =~ "PP-OCRv2" ]];then
inference_model_url=https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/${model}.tar inference_model_url=https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/${model}.tar
...@@ -146,48 +181,62 @@ for model in ${lite_model_list[*]}; do ...@@ -146,48 +181,62 @@ for model in ${lite_model_list[*]}; do
done done
``` ```
3.准备测试数据 4.准备测试数据
``` ```
data_url=https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/icdar2015_lite.tar data_url=https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/icdar2015_lite.tar
model_path=./inference_models
inference_model=${inference_model_url##*/}
data_file=${data_url##*/} data_file=${data_url##*/}
wget -nc -P ./inference_models ${inference_model_url}
wget -nc -P ./test_data ${data_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 ../ cd ./test_data && tar -xf ${data_file} && rm ${data_file} && cd ../
``` ```
4.准备Lite预测环境，此处需要下载或者编译Paddle-Lite预测库。 5.准备Lite预测环境，此处需要下载或者编译Paddle-Lite预测库。
``` ```
# prepare lite env # prepare lite env
paddlelite_zipfile=$(echo $paddlelite_url | awk -F "/" '{print $NF}') if [[ ${paddlelite_library_source} = "download" ]]; then
paddlelite_file=${paddlelite_zipfile:0:${end_index}} paddlelite_library_zipfile=$(echo $paddlelite_library_url | awk -F "/" '{print $NF}')
wget ${paddlelite_url} && tar -xf ${paddlelite_zipfile} paddlelite_library_file=${paddlelite_library_zipfile:0:${end_index}}
mkdir -p ${paddlelite_file}/demo/cxx/ocr/test_lite wget ${paddlelite_library_url} && tar -xf ${paddlelite_library_zipfile}
cp -r ${model_path}/*_opt.nb test_data ${paddlelite_file}/demo/cxx/ocr/test_lite cd ${paddlelite_library_zipfile}
cp ppocr/utils/ppocr_keys_v1.txt deploy/lite/config.txt ${paddlelite_file}/demo/cxx/ocr/test_lite elif [[ ${paddlelite_library_source} = "compile" ]]; then
cp -r ./deploy/lite/* ${paddlelite_file}/demo/cxx/ocr/ git clone -b release/v2.10 https://github.com/PaddlePaddle/Paddle-Lite.git
cp ${paddlelite_file}/cxx/lib/libpaddle_light_api_shared.so ${paddlelite_file}/demo/cxx/ocr/test_lite cd Paddle-Lite
cp ${FILENAME} test_tipc/test_lite_arm_cpp.sh test_tipc/common_func.sh ${paddlelite_file}/demo/cxx/ocr/test_lite ./lite/tools/build_android.sh --arch=armv8 --with_cv=ON --with_extra=ON --toolchain=clang --with_opencl=${compile_with_opencl}
cd ${paddlelite_file}/demo/cxx/ocr/ cd ../
git clone https://github.com/cuicheng01/AutoLog.git 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
``` ```
5.交叉编译获得在手机上的可以运行的可执行文件
6.准备编译以及后续在Lite设备上测试需要的文件
``` ```
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
```
7.交叉编译获得在手机上的可以运行的可执行文件及删除当前目录下无用文件
```
make -j make -j
sleep 1 sleep 1
make -j make -j
cp ocr_db_crnn test_lite && cp test_lite/libpaddle_light_api_shared.so test_lite/libc++_shared.so 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} tar -cf test_lite.tar ./test_lite && cp test_lite.tar ${current_dir} && cd ${current_dir}
rm -rf ${paddlelite_file}* && rm -rf ${model_path} rm -rf ${paddlelite_library_file}* && rm -rf ${model_path}
``` ```
运行结果会在当前目录上生成test_lite.tar，里边的内容大致如下： 运行结果会在当前目录上生成test_lite.tar，里边的内容大致如下：
``` ```
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或