mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2025-10-19 06:54:41 +08:00
[Backend] A311D support (#825)
* add A311D support * update code * update toolchain * update opencv_armhf lib * update libs * update code * add install script * update bos link * update toolchain
This commit is contained in:
yeliang2258
@@ -41,12 +41,6 @@ elseif(IOS)
|
|||||||
else()
|
else()
|
||||||
if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
|
if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
|
||||||
set(OPENCV_FILENAME "opencv-linux-aarch64-3.4.14")
|
set(OPENCV_FILENAME "opencv-linux-aarch64-3.4.14")
|
||||||
else()
|
|
||||||
if(ENABLE_TIMVX)
|
|
||||||
set(OPENCV_FILENAME "opencv-armv7hf")
|
|
||||||
else()
|
|
||||||
set(OPENCV_FILENAME "opencv-linux-x64-3.4.16")
|
|
||||||
endif()
|
|
||||||
endif()
|
endif()
|
||||||
if(ENABLE_OPENCV_CUDA)
|
if(ENABLE_OPENCV_CUDA)
|
||||||
if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
|
if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
|
||||||
@@ -56,15 +50,20 @@ else()
|
|||||||
endif()
|
endif()
|
||||||
endif()
|
endif()
|
||||||
|
|
||||||
|
if(NOT OPENCV_FILENAME)
|
||||||
|
set(OPENCV_FILENAME "opencv-linux-x64-3.4.16")
|
||||||
|
endif()
|
||||||
|
|
||||||
set(OPENCV_INSTALL_DIR ${THIRD_PARTY_PATH}/install/)
|
set(OPENCV_INSTALL_DIR ${THIRD_PARTY_PATH}/install/)
|
||||||
if(ANDROID)
|
if(ANDROID)
|
||||||
set(OPENCV_URL_PREFIX "https://bj.bcebos.com/fastdeploy/third_libs")
|
set(OPENCV_URL_PREFIX "https://bj.bcebos.com/fastdeploy/third_libs")
|
||||||
elseif(ENABLE_TIMVX)
|
|
||||||
set(OPENCV_URL_PREFIX "https://bj.bcebos.com/fastdeploy/test")
|
|
||||||
else() # TODO: use fastdeploy/third_libs instead.
|
else() # TODO: use fastdeploy/third_libs instead.
|
||||||
set(OPENCV_URL_PREFIX "https://bj.bcebos.com/paddle2onnx/libs")
|
set(OPENCV_URL_PREFIX "https://bj.bcebos.com/paddle2onnx/libs")
|
||||||
endif()
|
endif()
|
||||||
set(OPENCV_URL ${OPENCV_URL_PREFIX}/${OPENCV_FILENAME}${COMPRESSED_SUFFIX})
|
if(NOT OPENCV_URL)
|
||||||
|
set(OPENCV_URL ${OPENCV_URL_PREFIX}/${OPENCV_FILENAME}${COMPRESSED_SUFFIX})
|
||||||
|
endif()
|
||||||
|
|
||||||
|
|
||||||
if(BUILD_ON_JETSON)
|
if(BUILD_ON_JETSON)
|
||||||
if(EXISTS /usr/lib/aarch64-linux-gnu/cmake/opencv4/)
|
if(EXISTS /usr/lib/aarch64-linux-gnu/cmake/opencv4/)
|
||||||
@@ -186,9 +185,8 @@ else()
|
|||||||
endif()
|
endif()
|
||||||
file(RENAME ${THIRD_PARTY_PATH}/install/${OPENCV_FILENAME}/ ${THIRD_PARTY_PATH}/install/opencv)
|
file(RENAME ${THIRD_PARTY_PATH}/install/${OPENCV_FILENAME}/ ${THIRD_PARTY_PATH}/install/opencv)
|
||||||
set(OPENCV_FILENAME opencv)
|
set(OPENCV_FILENAME opencv)
|
||||||
set(OpenCV_DIR ${THIRD_PARTY_PATH}/install/${OPENCV_FILENAME})
|
if(NOT OpenCV_DIR)
|
||||||
if(ENABLE_TIMVX)
|
set(OpenCV_DIR ${THIRD_PARTY_PATH}/install/${OPENCV_FILENAME})
|
||||||
set(OpenCV_DIR ${OpenCV_DIR}/lib/cmake/opencv4)
|
|
||||||
endif()
|
endif()
|
||||||
if (WIN32)
|
if (WIN32)
|
||||||
set(OpenCV_DIR ${OpenCV_DIR}/build)
|
set(OpenCV_DIR ${OpenCV_DIR}/build)
|
||||||
|
|||||||
@@ -49,22 +49,20 @@ if(ANDROID)
|
|||||||
endif()
|
endif()
|
||||||
endif()
|
endif()
|
||||||
|
|
||||||
if(WIN32 OR APPLE OR IOS)
|
if(NOT PADDLELITE_URL)
|
||||||
message(FATAL_ERROR "Doesn't support windows/mac/ios platform with backend Paddle Lite now.")
|
if(WIN32 OR APPLE OR IOS)
|
||||||
elseif(ANDROID)
|
message(FATAL_ERROR "Doesn't support windows/mac/ios platform with backend Paddle Lite now.")
|
||||||
set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-android-${ANDROID_ABI}-latest-dev.tgz")
|
elseif(ANDROID)
|
||||||
if(ANDROID_ABI MATCHES "arm64-v8a")
|
set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-android-${ANDROID_ABI}-latest-dev.tgz")
|
||||||
set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-android-${ANDROID_ABI}-fp16-latest-dev.tgz")
|
if(ANDROID_ABI MATCHES "arm64-v8a")
|
||||||
endif()
|
set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-android-${ANDROID_ABI}-fp16-latest-dev.tgz")
|
||||||
else() # Linux
|
endif()
|
||||||
if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
|
else() # Linux
|
||||||
set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-linux-arm64-20221209.tgz")
|
if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
|
||||||
else()
|
set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-linux-arm64-20221209.tgz")
|
||||||
if(ENABLE_TIMVX)
|
|
||||||
set(PADDLELITE_URL "https://bj.bcebos.com/fastdeploy/test/lite-linux_armhf_1130.tgz")
|
|
||||||
else()
|
else()
|
||||||
message(FATAL_ERROR "Only support Linux aarch64 or ENABLE_TIMVX now, x64 is not supported with backend Paddle Lite.")
|
message(FATAL_ERROR "Only support Linux aarch64 now, x64 is not supported with backend Paddle Lite.")
|
||||||
endif()
|
endif()
|
||||||
endif()
|
endif()
|
||||||
endif()
|
endif()
|
||||||
|
|
||||||
|
|||||||
@@ -1,54 +1,45 @@
|
|||||||
if (NOT DEFINED CMAKE_SYSTEM_PROCESSOR)
|
|
||||||
set(CMAKE_SYSTEM_NAME Linux)
|
|
||||||
set(CMAKE_SYSTEM_PROCESSOR arm)
|
|
||||||
set(CMAKE_C_COMPILER "arm-linux-gnueabihf-gcc")
|
|
||||||
set(CMAKE_CXX_COMPILER "arm-linux-gnueabihf-g++")
|
|
||||||
set(CMAKE_CXX_FLAGS "-march=armv7-a -mfloat-abi=hard -mfpu=neon-vfpv4 ${CMAKE_CXX_FLAGS}")
|
|
||||||
set(CMAKE_C_FLAGS "-march=armv7-a -mfloat-abi=hard -mfpu=neon-vfpv4 ${CMAKE_C_FLAGS}" )
|
|
||||||
set(CMAKE_BUILD_TYPE MinSizeRel)
|
|
||||||
else()
|
|
||||||
if(NOT ${ENABLE_LITE_BACKEND})
|
|
||||||
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_LITE_BACKEND=ON")
|
|
||||||
set(ENABLE_LITE_BACKEND ON)
|
|
||||||
endif()
|
|
||||||
if(${ENABLE_PADDLE_FRONTEND})
|
|
||||||
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_PADDLE_FRONTEND=OFF")
|
|
||||||
set(ENABLE_PADDLE_FRONTEND OFF)
|
|
||||||
endif()
|
|
||||||
if(${ENABLE_ORT_BACKEND})
|
|
||||||
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_ORT_BACKEND=OFF")
|
|
||||||
set(ENABLE_ORT_BACKEND OFF)
|
|
||||||
endif()
|
|
||||||
if(${ENABLE_PADDLE_BACKEND})
|
|
||||||
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_PADDLE_BACKEND=OFF")
|
|
||||||
set(ENABLE_PADDLE_BACKEND OFF)
|
|
||||||
endif()
|
|
||||||
if(${ENABLE_OPENVINO_BACKEND})
|
|
||||||
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_OPENVINO_BACKEND=OFF")
|
|
||||||
set(ENABLE_OPENVINO_BACKEND OFF)
|
|
||||||
endif()
|
|
||||||
if(${ENABLE_TRT_BACKEND})
|
|
||||||
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_TRT_BACKEND=OFF")
|
|
||||||
set(ENABLE_TRT_BACKEND OFF)
|
|
||||||
endif()
|
|
||||||
|
|
||||||
if(${WITH_GPU})
|
if(NOT ${ENABLE_LITE_BACKEND})
|
||||||
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DWITH_GPU=OFF")
|
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_LITE_BACKEND=ON")
|
||||||
set(WITH_GPU OFF)
|
set(ENABLE_LITE_BACKEND ON)
|
||||||
endif()
|
endif()
|
||||||
|
if(${ENABLE_PADDLE_FRONTEND})
|
||||||
if(${ENABLE_OPENCV_CUDA})
|
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_PADDLE_FRONTEND=OFF")
|
||||||
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_OPENCV_CUDA=OFF")
|
set(ENABLE_PADDLE_FRONTEND OFF)
|
||||||
set(ENABLE_OPENCV_CUDA OFF)
|
endif()
|
||||||
endif()
|
if(${ENABLE_ORT_BACKEND})
|
||||||
|
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_ORT_BACKEND=OFF")
|
||||||
if(${ENABLE_TEXT})
|
set(ENABLE_ORT_BACKEND OFF)
|
||||||
set(ENABLE_TEXT OFF CACHE BOOL "Force ENABLE_TEXT OFF" FORCE)
|
endif()
|
||||||
message(STATUS "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_TEXT=OFF")
|
if(${ENABLE_PADDLE_BACKEND})
|
||||||
endif()
|
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_PADDLE_BACKEND=OFF")
|
||||||
if (DEFINED CMAKE_INSTALL_PREFIX)
|
set(ENABLE_PADDLE_BACKEND OFF)
|
||||||
install(FILES ${PROJECT_SOURCE_DIR}/cmake/timvx.cmake DESTINATION ${CMAKE_INSTALL_PREFIX})
|
endif()
|
||||||
endif()
|
if(${ENABLE_OPENVINO_BACKEND})
|
||||||
|
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_OPENVINO_BACKEND=OFF")
|
||||||
|
set(ENABLE_OPENVINO_BACKEND OFF)
|
||||||
|
endif()
|
||||||
|
if(${ENABLE_TRT_BACKEND})
|
||||||
|
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_TRT_BACKEND=OFF")
|
||||||
|
set(ENABLE_TRT_BACKEND OFF)
|
||||||
endif()
|
endif()
|
||||||
|
|
||||||
|
if(${WITH_GPU})
|
||||||
|
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DWITH_GPU=OFF")
|
||||||
|
set(WITH_GPU OFF)
|
||||||
|
endif()
|
||||||
|
|
||||||
|
if(${ENABLE_OPENCV_CUDA})
|
||||||
|
message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_OPENCV_CUDA=OFF")
|
||||||
|
set(ENABLE_OPENCV_CUDA OFF)
|
||||||
|
endif()
|
||||||
|
|
||||||
|
if(${ENABLE_TEXT})
|
||||||
|
set(ENABLE_TEXT OFF CACHE BOOL "Force ENABLE_TEXT OFF" FORCE)
|
||||||
|
message(STATUS "While compiling with -DWITH_TIMVX=ON, will force to set -DENABLE_TEXT=OFF")
|
||||||
|
endif()
|
||||||
|
|
||||||
|
install(FILES ${PROJECT_SOURCE_DIR}/cmake/timvx.cmake DESTINATION ${CMAKE_INSTALL_PREFIX})
|
||||||
|
install(FILES ${PROJECT_SOURCE_DIR}/cmake/toolchain.cmake DESTINATION ${CMAKE_INSTALL_PREFIX})
|
||||||
|
|
||||||
|
|
||||||
|
|||||||
38
cmake/toolchain.cmake
Executable file
38
cmake/toolchain.cmake
Executable file
@@ -0,0 +1,38 @@
|
|||||||
|
if (DEFINED TARGET_ABI)
|
||||||
|
set(CMAKE_SYSTEM_NAME Linux)
|
||||||
|
set(CMAKE_BUILD_TYPE MinSizeRel)
|
||||||
|
if(${TARGET_ABI} MATCHES "armhf")
|
||||||
|
set(CMAKE_SYSTEM_PROCESSOR arm)
|
||||||
|
set(CMAKE_C_COMPILER "arm-linux-gnueabihf-gcc")
|
||||||
|
set(CMAKE_CXX_COMPILER "arm-linux-gnueabihf-g++")
|
||||||
|
set(CMAKE_CXX_FLAGS "-march=armv7-a -mfloat-abi=hard -mfpu=neon-vfpv4 ${CMAKE_CXX_FLAGS}")
|
||||||
|
set(CMAKE_C_FLAGS "-march=armv7-a -mfloat-abi=hard -mfpu=neon-vfpv4 ${CMAKE_C_FLAGS}" )
|
||||||
|
set(OPENCV_URL "https://bj.bcebos.com/fastdeploy/third_libs/opencv-linux-armv7hf-4.6.0.tgz")
|
||||||
|
set(OPENCV_FILENAME "opencv-linux-armv7hf-4.6.0")
|
||||||
|
if(WITH_TIMVX)
|
||||||
|
set(PADDLELITE_URL "https://bj.bcebos.com/fastdeploy/third_libs/lite-linux-armhf-timvx-1130.tgz")
|
||||||
|
else()
|
||||||
|
message(STATUS "PADDLELITE_URL will be configured if WITH_TIMVX=ON.")
|
||||||
|
endif()
|
||||||
|
set(THIRD_PARTY_PATH ${CMAKE_CURRENT_BINARY_DIR}/third_libs)
|
||||||
|
set(OpenCV_DIR ${THIRD_PARTY_PATH}/install/opencv/lib/cmake/opencv4)
|
||||||
|
elseif(${TARGET_ABI} MATCHES "arm64")
|
||||||
|
set(CMAKE_SYSTEM_PROCESSOR aarch64)
|
||||||
|
set(CMAKE_C_COMPILER "aarch64-linux-gnu-gcc")
|
||||||
|
set(CMAKE_CXX_COMPILER "aarch64-linux-gnu-g++")
|
||||||
|
set(CMAKE_CXX_FLAGS "-march=armv8-a ${CMAKE_CXX_FLAGS}")
|
||||||
|
set(CMAKE_C_FLAGS "-march=armv8-a ${CMAKE_C_FLAGS}")
|
||||||
|
set(OPENCV_URL "https://bj.bcebos.com/fastdeploy/third_libs/opencv-linux-aarch64-4.6.0.tgz")
|
||||||
|
set(OPENCV_FILENAME "opencv-linux-aarch64-4.6.0")
|
||||||
|
if(WITH_TIMVX)
|
||||||
|
set(PADDLELITE_URL "https://bj.bcebos.com/fastdeploy/third_libs/lite-linux-aarch64-timvx-20221209.tgz")
|
||||||
|
else()
|
||||||
|
set(PADDLELITE_URL "https://bj.bcebos.com/fastdeploy/third_libs/lite-linux-arm64-20221209.tgz")
|
||||||
|
endif()
|
||||||
|
set(THIRD_PARTY_PATH ${CMAKE_CURRENT_BINARY_DIR}/third_libs)
|
||||||
|
set(OpenCV_DIR ${THIRD_PARTY_PATH}/install/opencv/lib/cmake/opencv4)
|
||||||
|
else()
|
||||||
|
message(FATAL_ERROR "When cross-compiling, please set the -DTARGET_ABI to arm64 or armhf.")
|
||||||
|
endif()
|
||||||
|
endif()
|
||||||
|
|
||||||
@@ -12,6 +12,7 @@
|
|||||||
- [Jetson部署环境](jetson.md)
|
- [Jetson部署环境](jetson.md)
|
||||||
- [Android平台部署环境](android.md)
|
- [Android平台部署环境](android.md)
|
||||||
- [瑞芯微RV1126部署环境](rv1126.md)
|
- [瑞芯微RV1126部署环境](rv1126.md)
|
||||||
|
- [晶晨A311D部署环境](a311d.md)
|
||||||
|
|
||||||
|
|
||||||
## FastDeploy编译选项说明
|
## FastDeploy编译选项说明
|
||||||
@@ -22,7 +23,7 @@
|
|||||||
| ENABLE_PADDLE_BACKEND | 默认OFF，是否编译集成Paddle Inference后端(CPU/GPU上推荐打开) |
|
| ENABLE_PADDLE_BACKEND | 默认OFF，是否编译集成Paddle Inference后端(CPU/GPU上推荐打开) |
|
||||||
| ENABLE_LITE_BACKEND | 默认OFF，是否编译集成Paddle Lite后端(编译Android库时需要设置为ON) |
|
| ENABLE_LITE_BACKEND | 默认OFF，是否编译集成Paddle Lite后端(编译Android库时需要设置为ON) |
|
||||||
| ENABLE_RKNPU2_BACKEND | 默认OFF，是否编译集成RKNPU2后端(RK3588/RK3568/RK3566上推荐打开) |
|
| ENABLE_RKNPU2_BACKEND | 默认OFF，是否编译集成RKNPU2后端(RK3588/RK3568/RK3566上推荐打开) |
|
||||||
| ENABLE_TIMVX | 默认OFF，需要在RV1126/RV1109上部署时，需设置为ON |
|
| WITH_TIMVX | 默认OFF，需要在RV1126/RV1109/A311D上部署时，需设置为ON |
|
||||||
| ENABLE_TRT_BACKEND | 默认OFF，是否编译集成TensorRT后端(GPU上推荐打开) |
|
| ENABLE_TRT_BACKEND | 默认OFF，是否编译集成TensorRT后端(GPU上推荐打开) |
|
||||||
| ENABLE_OPENVINO_BACKEND | 默认OFF，是否编译集成OpenVINO后端(CPU上推荐打开) |
|
| ENABLE_OPENVINO_BACKEND | 默认OFF，是否编译集成OpenVINO后端(CPU上推荐打开) |
|
||||||
| ENABLE_VISION | 默认OFF，是否编译集成视觉模型的部署模块 |
|
| ENABLE_VISION | 默认OFF，是否编译集成视觉模型的部署模块 |
|
||||||
|
|||||||
107
docs/cn/build_and_install/a311d.md
Executable file
107
docs/cn/build_and_install/a311d.md
Executable file
@@ -0,0 +1,107 @@
|
|||||||
|
# 晶晨 A311D 部署环境编译安装
|
||||||
|
|
||||||
|
FastDeploy 基于 Paddle-Lite 后端支持在晶晨 NPU 上进行部署推理。
|
||||||
|
更多详细的信息请参考：[PaddleLite部署示例](https://www.paddlepaddle.org.cn/lite/develop/demo_guides/verisilicon_timvx.html)。
|
||||||
|
|
||||||
|
本文档介绍如何编译基于 PaddleLite 的 C++ FastDeploy 交叉编译库。
|
||||||
|
|
||||||
|
相关编译选项说明如下：
|
||||||
|
|编译选项|默认值|说明|备注|
|
||||||
|
|:---|:---|:---|:---|
|
||||||
|
|ENABLE_LITE_BACKEND|OFF|编译A311D部署库时需要设置为ON| - |
|
||||||
|
|WITH_TIMVX|OFF|编译A311D部署库时需要设置为ON| - |
|
||||||
|
|
||||||
|
更多编译选项请参考[FastDeploy编译选项说明](./README.md)
|
||||||
|
|
||||||
|
## 交叉编译环境搭建
|
||||||
|
|
||||||
|
### 宿主机环境需求
|
||||||
|
- os：Ubuntu == 16.04
|
||||||
|
- cmake： version >= 3.10.0
|
||||||
|
|
||||||
|
### 环境搭建
|
||||||
|
可以进入 FastDeploy/tools/timvx 目录，使用如下命令一键安装：
|
||||||
|
```bash
|
||||||
|
cd FastDeploy/tools/timvx
|
||||||
|
bash install.sh
|
||||||
|
```
|
||||||
|
也可以按照如下命令安装：
|
||||||
|
```bash
|
||||||
|
# 1. Install basic software
|
||||||
|
apt update
|
||||||
|
apt-get install -y --no-install-recommends \
|
||||||
|
gcc g++ git make wget python unzip
|
||||||
|
|
||||||
|
# 2. Install arm gcc toolchains
|
||||||
|
apt-get install -y --no-install-recommends \
|
||||||
|
g++-arm-linux-gnueabi gcc-arm-linux-gnueabi \
|
||||||
|
g++-arm-linux-gnueabihf gcc-arm-linux-gnueabihf \
|
||||||
|
gcc-aarch64-linux-gnu g++-aarch64-linux-gnu
|
||||||
|
|
||||||
|
# 3. Install cmake 3.10 or above
|
||||||
|
wget -c https://mms-res.cdn.bcebos.com/cmake-3.10.3-Linux-x86_64.tar.gz && \
|
||||||
|
tar xzf cmake-3.10.3-Linux-x86_64.tar.gz && \
|
||||||
|
mv cmake-3.10.3-Linux-x86_64 /opt/cmake-3.10 && \
|
||||||
|
ln -s /opt/cmake-3.10/bin/cmake /usr/bin/cmake && \
|
||||||
|
ln -s /opt/cmake-3.10/bin/ccmake /usr/bin/ccmake
|
||||||
|
```
|
||||||
|
|
||||||
|
## 基于 PaddleLite 的 FastDeploy 交叉编译库编译
|
||||||
|
搭建好交叉编译环境之后，编译命令如下：
|
||||||
|
```bash
|
||||||
|
# Download the latest source code
|
||||||
|
git clone https://github.com/PaddlePaddle/FastDeploy.git
|
||||||
|
cd FastDeploy
|
||||||
|
mkdir build && cd build
|
||||||
|
|
||||||
|
# CMake configuration with A311D toolchain
|
||||||
|
cmake -DCMAKE_TOOLCHAIN_FILE=./../cmake/toolchain.cmake \
|
||||||
|
-DWITH_TIMVX=ON \
|
||||||
|
-DTARGET_ABI=arm64 \
|
||||||
|
-DCMAKE_INSTALL_PREFIX=fastdeploy-tmivx \
|
||||||
|
-DENABLE_VISION=ON \ # 是否编译集成视觉模型的部署模块，可选择开启
|
||||||
|
-Wno-dev ..
|
||||||
|
|
||||||
|
# Build FastDeploy A311D C++ SDK
|
||||||
|
make -j8
|
||||||
|
make install
|
||||||
|
```
|
||||||
|
编译完成之后，会生成 fastdeploy-tmivx 目录，表示基于 PadddleLite TIM-VX 的 FastDeploy 库编译完成。
|
||||||
|
|
||||||
|
## 准备设备运行环境
|
||||||
|
部署前要保证晶晨 Linux Kernel NPU 驱动 galcore.so 版本及所适用的芯片型号与依赖库保持一致，在部署前，请登录开发板，并通过命令行输入以下命令查询 NPU 驱动版本，晶晨建议的驱动版本为：6.4.4.3
|
||||||
|
```bash
|
||||||
|
dmesg | grep Galcore
|
||||||
|
```
|
||||||
|
|
||||||
|
如果当前版本不符合上述，请用户仔细阅读以下内容，以保证底层 NPU 驱动环境正确。
|
||||||
|
|
||||||
|
有两种方式可以修改当前的 NPU 驱动版本：
|
||||||
|
1. 手动替换 NPU 驱动版本。（推荐）
|
||||||
|
2. 刷机，刷取 NPU 驱动版本符合要求的固件。
|
||||||
|
|
||||||
|
### 手动替换 NPU 驱动版本
|
||||||
|
1. 使用如下命令下载解压 PaddleLite demo，其中提供了现成的驱动文件
|
||||||
|
```bash
|
||||||
|
wget https://paddlelite-demo.bj.bcebos.com/devices/generic/PaddleLite-generic-demo.tar.gz
|
||||||
|
tar -xf PaddleLite-generic-demo.tar.gz
|
||||||
|
```
|
||||||
|
2. 使用 `uname -a` 查看 `Linux Kernel` 版本，确定为 `Linux` 系统 4.19.111 版本，
|
||||||
|
3. 将 `PaddleLite-generic-demo/libs/PaddleLite/linux/arm64/lib/verisilicon_timvx/viv_sdk_6_4_4_3/lib/a311d/4.9.113` 路径下的 `galcore.ko` 上传至开发板。
|
||||||
|
|
||||||
|
4. 登录开发板，命令行输入 `sudo rmmod galcore` 来卸载原始驱动，输入 `sudo insmod galcore.ko` 来加载传上设备的驱动。（是否需要 sudo 根据开发板实际情况，部分 adb 链接的设备请提前 adb root）。此步骤如果操作失败，请跳转至方法 2。
|
||||||
|
5. 在开发板中输入 `dmesg | grep Galcore` 查询 NPU 驱动版本，确定为：6.4.4.3
|
||||||
|
|
||||||
|
### 刷机
|
||||||
|
根据具体的开发板型号，向开发板卖家或官网客服索要 6.4.4.3 版本 NPU 驱动对应的固件和刷机方法。
|
||||||
|
|
||||||
|
更多细节请参考：[PaddleLite准备设备环境](https://www.paddlepaddle.org.cn/lite/develop/demo_guides/verisilicon_timvx.html#zhunbeishebeihuanjing)
|
||||||
|
|
||||||
|
## 基于 FastDeploy 在 A311D 上的部署示例
|
||||||
|
1. A311D 上部署 PaddleClas 分类模型请参考：[PaddleClas 分类模型在 A311D 上的 C++ 部署示例](../../../examples/vision/classification/paddleclas/a311d/README.md)
|
||||||
|
|
||||||
|
2. A311D 上部署 PPYOLOE 检测模型请参考：[PPYOLOE 检测模型在 A311D 上的 C++ 部署示例](../../../examples/vision/detection/paddledetection/a311d/README.md)
|
||||||
|
|
||||||
|
3. A311D 上部署 YOLOv5 检测模型请参考：[YOLOv5 检测模型在 A311D 上的 C++ 部署示例](../../../examples/vision/detection/yolov5/a311d/README.md)
|
||||||
|
|
||||||
|
4. A311D 上部署 PP-LiteSeg 分割模型请参考：[PP-LiteSeg 分割模型在 A311D 上的 C++ 部署示例](../../../examples/vision/segmentation/paddleseg/a311d/README.md)
|
||||||
@@ -9,7 +9,7 @@ FastDeploy基于 Paddle-Lite 后端支持在瑞芯微（Rockchip）Soc 上进行
|
|||||||
|编译选项|默认值|说明|备注|
|
|编译选项|默认值|说明|备注|
|
||||||
|:---|:---|:---|:---|
|
|:---|:---|:---|:---|
|
||||||
|ENABLE_LITE_BACKEND|OFF|编译RK库时需要设置为ON| - |
|
|ENABLE_LITE_BACKEND|OFF|编译RK库时需要设置为ON| - |
|
||||||
|ENABLE_TIMVX|OFF|编译RK库时需要设置为ON| - |
|
|WITH_TIMVX|OFF|编译RK库时需要设置为ON| - |
|
||||||
|
|
||||||
更多编译选项请参考[FastDeploy编译选项说明](./README.md)
|
更多编译选项请参考[FastDeploy编译选项说明](./README.md)
|
||||||
|
|
||||||
@@ -20,6 +20,12 @@ FastDeploy基于 Paddle-Lite 后端支持在瑞芯微（Rockchip）Soc 上进行
|
|||||||
- cmake： version >= 3.10.0
|
- cmake： version >= 3.10.0
|
||||||
|
|
||||||
### 环境搭建
|
### 环境搭建
|
||||||
|
可以进入 FastDeploy/tools/timvx 目录，使用如下命令一键安装：
|
||||||
|
```bash
|
||||||
|
cd FastDeploy/tools/timvx
|
||||||
|
bash install.sh
|
||||||
|
```
|
||||||
|
也可以按照如下命令安装：
|
||||||
```bash
|
```bash
|
||||||
# 1. Install basic software
|
# 1. Install basic software
|
||||||
apt update
|
apt update
|
||||||
@@ -49,8 +55,9 @@ cd FastDeploy
|
|||||||
mkdir build && cd build
|
mkdir build && cd build
|
||||||
|
|
||||||
# CMake configuration with RK toolchain
|
# CMake configuration with RK toolchain
|
||||||
cmake -DCMAKE_TOOLCHAIN_FILE=./../cmake/timvx.cmake \
|
cmake -DCMAKE_TOOLCHAIN_FILE=./../cmake/toolchain.cmake \
|
||||||
-DENABLE_TIMVX=ON \
|
-DWITH_TIMVX=ON \
|
||||||
|
-DTARGET_ABI=armhf \
|
||||||
-DCMAKE_INSTALL_PREFIX=fastdeploy-tmivx \
|
-DCMAKE_INSTALL_PREFIX=fastdeploy-tmivx \
|
||||||
-DENABLE_VISION=ON \ # 是否编译集成视觉模型的部署模块，可选择开启
|
-DENABLE_VISION=ON \ # 是否编译集成视觉模型的部署模块，可选择开启
|
||||||
-Wno-dev ..
|
-Wno-dev ..
|
||||||
|
|||||||
11
examples/vision/classification/paddleclas/a311d/README.md
Executable file
11
examples/vision/classification/paddleclas/a311d/README.md
Executable file
@@ -0,0 +1,11 @@
|
|||||||
|
# PaddleClas 量化模型在 A311D 上的部署
|
||||||
|
目前 FastDeploy 已经支持基于 PaddleLite 部署 PaddleClas 量化模型到 A311D 上。
|
||||||
|
|
||||||
|
模型的量化和量化模型的下载请参考：[模型量化](../quantize/README.md)
|
||||||
|
|
||||||
|
|
||||||
|
## 详细部署文档
|
||||||
|
|
||||||
|
在 A311D 上只支持 C++ 的部署。
|
||||||
|
|
||||||
|
- [C++部署](cpp)
|
||||||
38
examples/vision/classification/paddleclas/a311d/cpp/CMakeLists.txt
Executable file
38
examples/vision/classification/paddleclas/a311d/cpp/CMakeLists.txt
Executable file
@@ -0,0 +1,38 @@
|
|||||||
|
PROJECT(infer_demo C CXX)
|
||||||
|
CMAKE_MINIMUM_REQUIRED (VERSION 3.10)
|
||||||
|
|
||||||
|
# 指定下载解压后的fastdeploy库路径
|
||||||
|
option(FASTDEPLOY_INSTALL_DIR "Path of downloaded fastdeploy sdk.")
|
||||||
|
|
||||||
|
include(${FASTDEPLOY_INSTALL_DIR}/FastDeploy.cmake)
|
||||||
|
|
||||||
|
# 添加FastDeploy依赖头文件
|
||||||
|
include_directories(${FASTDEPLOY_INCS})
|
||||||
|
include_directories(${FastDeploy_INCLUDE_DIRS})
|
||||||
|
|
||||||
|
add_executable(infer_demo ${PROJECT_SOURCE_DIR}/infer.cc)
|
||||||
|
# 添加FastDeploy库依赖
|
||||||
|
target_link_libraries(infer_demo ${FASTDEPLOY_LIBS})
|
||||||
|
|
||||||
|
set(CMAKE_INSTALL_PREFIX ${CMAKE_SOURCE_DIR}/build/install)
|
||||||
|
|
||||||
|
install(TARGETS infer_demo DESTINATION ./)
|
||||||
|
|
||||||
|
install(DIRECTORY models DESTINATION ./)
|
||||||
|
install(DIRECTORY images DESTINATION ./)
|
||||||
|
# install(DIRECTORY run_with_adb.sh DESTINATION ./)
|
||||||
|
|
||||||
|
file(GLOB FASTDEPLOY_LIBS ${FASTDEPLOY_INSTALL_DIR}/lib/*)
|
||||||
|
install(PROGRAMS ${FASTDEPLOY_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB OPENCV_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/opencv/lib/lib*)
|
||||||
|
install(PROGRAMS ${OPENCV_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB PADDLELITE_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/paddlelite/lib/lib*)
|
||||||
|
install(PROGRAMS ${PADDLELITE_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB TIMVX_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/paddlelite/lib/verisilicon_timvx/*)
|
||||||
|
install(PROGRAMS ${TIMVX_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB ADB_TOOLS run_with_adb.sh)
|
||||||
|
install(PROGRAMS ${ADB_TOOLS} DESTINATION ./)
|
||||||
53
examples/vision/classification/paddleclas/a311d/cpp/README.md
Executable file
53
examples/vision/classification/paddleclas/a311d/cpp/README.md
Executable file
@@ -0,0 +1,53 @@
|
|||||||
|
# PaddleClas A311D 开发板 C++ 部署示例
|
||||||
|
本目录下提供的 `infer.cc`，可以帮助用户快速完成 PaddleClas 量化模型在 A311D 上的部署推理加速。
|
||||||
|
|
||||||
|
## 部署准备
|
||||||
|
### FastDeploy 交叉编译环境准备
|
||||||
|
- 1. 软硬件环境满足要求，以及交叉编译环境的准备，请参考：[FastDeploy 交叉编译环境准备](../../../../../../docs/cn/build_and_install/a311d.md#交叉编译环境搭建)
|
||||||
|
|
||||||
|
### 量化模型准备
|
||||||
|
- 1. 用户可以直接使用由 FastDeploy 提供的量化模型进行部署。
|
||||||
|
- 2. 用户可以使用 FastDeploy 提供的[一键模型自动化压缩工具](../../../../../../tools/common_tools/auto_compression/)，自行进行模型量化, 并使用产出的量化模型进行部署。(注意: 推理量化后的分类模型仍然需要FP32模型文件夹下的inference_cls.yaml文件, 自行量化的模型文件夹内不包含此 yaml 文件, 用户从 FP32 模型文件夹下复制此 yaml 文件到量化后的模型文件夹内即可.)
|
||||||
|
- 更多量化相关相关信息可查阅[模型量化](../../quantize/README.md)
|
||||||
|
|
||||||
|
## 在 A311D 上部署量化后的 ResNet50_Vd 分类模型
|
||||||
|
请按照以下步骤完成在 A311D 上部署 ResNet50_Vd 量化模型：
|
||||||
|
1. 交叉编译编译 FastDeploy 库，具体请参考：[交叉编译 FastDeploy](../../../../../../docs/cn/build_and_install/a311d.md#基于-paddlelite-的-fastdeploy-交叉编译库编译)
|
||||||
|
|
||||||
|
2. 将编译后的库拷贝到当前目录，可使用如下命令：
|
||||||
|
```bash
|
||||||
|
cp -r FastDeploy/build/fastdeploy-tmivx/ FastDeploy/examples/vision/classification/paddleclas/a311d/cpp/
|
||||||
|
```
|
||||||
|
|
||||||
|
3. 在当前路径下载部署所需的模型和示例图片：
|
||||||
|
```bash
|
||||||
|
mkdir models && mkdir images
|
||||||
|
wget https://bj.bcebos.com/paddlehub/fastdeploy/ResNet50_vd_infer.tgz
|
||||||
|
tar -xvf ResNet50_vd_infer.tgz
|
||||||
|
cp -r ResNet50_vd_infer models
|
||||||
|
wget https://gitee.com/paddlepaddle/PaddleClas/raw/release/2.4/deploy/images/ImageNet/ILSVRC2012_val_00000010.jpeg
|
||||||
|
cp -r ILSVRC2012_val_00000010.jpeg images
|
||||||
|
```
|
||||||
|
|
||||||
|
4. 编译部署示例，可使入如下命令：
|
||||||
|
```bash
|
||||||
|
mkdir build && cd build
|
||||||
|
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=arm64 ..
|
||||||
|
make -j8
|
||||||
|
make install
|
||||||
|
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
||||||
|
```
|
||||||
|
|
||||||
|
5. 基于 adb 工具部署 ResNet50_vd 分类模型到晶晨 A311D，可使用如下命令：
|
||||||
|
```bash
|
||||||
|
# 进入 install 目录
|
||||||
|
cd FastDeploy/examples/vision/classification/paddleclas/a311d/cpp/build/install/
|
||||||
|
# 如下命令表示：bash run_with_adb.sh 需要运行的demo 模型路径 图片路径 设备的DEVICE_ID
|
||||||
|
bash run_with_adb.sh infer_demo ResNet50_vd_infer ILSVRC2012_val_00000010.jpeg $DEVICE_ID
|
||||||
|
```
|
||||||
|
|
||||||
|
部署成功后运行结果如下：
|
||||||
|
|
||||||
|
<img width="640" src="https://user-images.githubusercontent.com/30516196/200767389-26519e50-9e4f-4fe1-8d52-260718f73476.png">
|
||||||
|
|
||||||
|
需要特别注意的是，在 A311D 上部署的模型需要是量化后的模型，模型的量化请参考：[模型量化](../../../../../../docs/cn/quantize.md)
|
||||||
60
examples/vision/classification/paddleclas/a311d/cpp/infer.cc
Executable file
60
examples/vision/classification/paddleclas/a311d/cpp/infer.cc
Executable file
@@ -0,0 +1,60 @@
|
|||||||
|
// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
|
||||||
|
//
|
||||||
|
// 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.
|
||||||
|
#include <string>
|
||||||
|
#include "fastdeploy/vision.h"
|
||||||
|
#ifdef WIN32
|
||||||
|
const char sep = '\\';
|
||||||
|
#else
|
||||||
|
const char sep = '/';
|
||||||
|
#endif
|
||||||
|
|
||||||
|
void InitAndInfer(const std::string& model_dir, const std::string& image_file) {
|
||||||
|
auto model_file = model_dir + sep + "inference.pdmodel";
|
||||||
|
auto params_file = model_dir + sep + "inference.pdiparams";
|
||||||
|
auto config_file = model_dir + sep + "inference_cls.yaml";
|
||||||
|
|
||||||
|
fastdeploy::RuntimeOption option;
|
||||||
|
option.UseTimVX();
|
||||||
|
|
||||||
|
auto model = fastdeploy::vision::classification::PaddleClasModel(
|
||||||
|
model_file, params_file, config_file, option);
|
||||||
|
|
||||||
|
assert(model.Initialized());
|
||||||
|
|
||||||
|
auto im = cv::imread(image_file);
|
||||||
|
|
||||||
|
fastdeploy::vision::ClassifyResult res;
|
||||||
|
if (!model.Predict(im, &res)) {
|
||||||
|
std::cerr << "Failed to predict." << std::endl;
|
||||||
|
return;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::cout << res.Str() << std::endl;
|
||||||
|
|
||||||
|
}
|
||||||
|
|
||||||
|
int main(int argc, char* argv[]) {
|
||||||
|
if (argc < 3) {
|
||||||
|
std::cout << "Usage: infer_demo path/to/quant_model "
|
||||||
|
"path/to/image "
|
||||||
|
"e.g ./infer_demo ./ResNet50_vd_quant ./test.jpeg"
|
||||||
|
<< std::endl;
|
||||||
|
return -1;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::string model_dir = argv[1];
|
||||||
|
std::string test_image = argv[2];
|
||||||
|
InitAndInfer(model_dir, test_image);
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
47
examples/vision/classification/paddleclas/a311d/cpp/run_with_adb.sh
Executable file
47
examples/vision/classification/paddleclas/a311d/cpp/run_with_adb.sh
Executable file
@@ -0,0 +1,47 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
HOST_SPACE=${PWD}
|
||||||
|
echo ${HOST_SPACE}
|
||||||
|
WORK_SPACE=/data/local/tmp/test
|
||||||
|
|
||||||
|
# The first parameter represents the demo name
|
||||||
|
DEMO_NAME=image_classification_demo
|
||||||
|
if [ -n "$1" ]; then
|
||||||
|
DEMO_NAME=$1
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The second parameter represents the model name
|
||||||
|
MODEL_NAME=mobilenet_v1_fp32_224
|
||||||
|
if [ -n "$2" ]; then
|
||||||
|
MODEL_NAME=$2
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The third parameter indicates the name of the image to be tested
|
||||||
|
IMAGE_NAME=0001.jpg
|
||||||
|
if [ -n "$3" ]; then
|
||||||
|
IMAGE_NAME=$3
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The fourth parameter represents the ID of the device
|
||||||
|
ADB_DEVICE_NAME=
|
||||||
|
if [ -n "$4" ]; then
|
||||||
|
ADB_DEVICE_NAME="-s $4"
|
||||||
|
fi
|
||||||
|
|
||||||
|
# Set the environment variables required during the running process
|
||||||
|
EXPORT_ENVIRONMENT_VARIABLES="export GLOG_v=5; export VIV_VX_ENABLE_GRAPH_TRANSFORM=-pcq:1; export VIV_VX_SET_PER_CHANNEL_ENTROPY=100; export TIMVX_BATCHNORM_FUSION_MAX_ALLOWED_QUANT_SCALE_DEVIATION=300000; export VSI_NN_LOG_LEVEL=5;"
|
||||||
|
|
||||||
|
EXPORT_ENVIRONMENT_VARIABLES="${EXPORT_ENVIRONMENT_VARIABLES}export LD_LIBRARY_PATH=${WORK_SPACE}/lib:\$LD_LIBRARY_PATH;"
|
||||||
|
|
||||||
|
# Please install adb, and DON'T run this in the docker.
|
||||||
|
set -e
|
||||||
|
adb $ADB_DEVICE_NAME shell "rm -rf $WORK_SPACE"
|
||||||
|
adb $ADB_DEVICE_NAME shell "mkdir -p $WORK_SPACE"
|
||||||
|
|
||||||
|
# Upload the demo, librarys, model and test images to the device
|
||||||
|
adb $ADB_DEVICE_NAME push ${HOST_SPACE}/lib $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push ${HOST_SPACE}/${DEMO_NAME} $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push models $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push images $WORK_SPACE
|
||||||
|
|
||||||
|
# Execute the deployment demo
|
||||||
|
adb $ADB_DEVICE_NAME shell "cd $WORK_SPACE; ${EXPORT_ENVIRONMENT_VARIABLES} chmod +x ./${DEMO_NAME}; ./${DEMO_NAME} ./models/${MODEL_NAME} ./images/$IMAGE_NAME"
|
||||||
@@ -32,7 +32,7 @@ cp -r ILSVRC2012_val_00000010.jpeg images
|
|||||||
4. 编译部署示例，可使入如下命令：
|
4. 编译部署示例，可使入如下命令：
|
||||||
```bash
|
```bash
|
||||||
mkdir build && cd build
|
mkdir build && cd build
|
||||||
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/timvx.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx ..
|
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=armhf ..
|
||||||
make -j8
|
make -j8
|
||||||
make install
|
make install
|
||||||
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
||||||
|
|||||||
@@ -48,7 +48,6 @@ int main(int argc, char* argv[]) {
|
|||||||
if (argc < 3) {
|
if (argc < 3) {
|
||||||
std::cout << "Usage: infer_demo path/to/quant_model "
|
std::cout << "Usage: infer_demo path/to/quant_model "
|
||||||
"path/to/image "
|
"path/to/image "
|
||||||
"run_option, "
|
|
||||||
"e.g ./infer_demo ./ResNet50_vd_quant ./test.jpeg"
|
"e.g ./infer_demo ./ResNet50_vd_quant ./test.jpeg"
|
||||||
<< std::endl;
|
<< std::endl;
|
||||||
return -1;
|
return -1;
|
||||||
|
|||||||
11
examples/vision/detection/paddledetection/a311d/README.md
Executable file
11
examples/vision/detection/paddledetection/a311d/README.md
Executable file
@@ -0,0 +1,11 @@
|
|||||||
|
# PP-YOLOE 量化模型在 A311D 上的部署
|
||||||
|
目前 FastDeploy 已经支持基于 PaddleLite 部署 PP-YOLOE 量化模型到 A311D 上。
|
||||||
|
|
||||||
|
模型的量化和量化模型的下载请参考：[模型量化](../quantize/README.md)
|
||||||
|
|
||||||
|
|
||||||
|
## 详细部署文档
|
||||||
|
|
||||||
|
在 A311D 上只支持 C++ 的部署。
|
||||||
|
|
||||||
|
- [C++部署](cpp)
|
||||||
38
examples/vision/detection/paddledetection/a311d/cpp/CMakeLists.txt
Executable file
38
examples/vision/detection/paddledetection/a311d/cpp/CMakeLists.txt
Executable file
@@ -0,0 +1,38 @@
|
|||||||
|
PROJECT(infer_demo C CXX)
|
||||||
|
CMAKE_MINIMUM_REQUIRED (VERSION 3.10)
|
||||||
|
|
||||||
|
# 指定下载解压后的fastdeploy库路径
|
||||||
|
option(FASTDEPLOY_INSTALL_DIR "Path of downloaded fastdeploy sdk.")
|
||||||
|
|
||||||
|
include(${FASTDEPLOY_INSTALL_DIR}/FastDeploy.cmake)
|
||||||
|
|
||||||
|
# 添加FastDeploy依赖头文件
|
||||||
|
include_directories(${FASTDEPLOY_INCS})
|
||||||
|
include_directories(${FastDeploy_INCLUDE_DIRS})
|
||||||
|
|
||||||
|
add_executable(infer_demo ${PROJECT_SOURCE_DIR}/infer_ppyoloe.cc)
|
||||||
|
# 添加FastDeploy库依赖
|
||||||
|
target_link_libraries(infer_demo ${FASTDEPLOY_LIBS})
|
||||||
|
|
||||||
|
set(CMAKE_INSTALL_PREFIX ${CMAKE_SOURCE_DIR}/build/install)
|
||||||
|
|
||||||
|
install(TARGETS infer_demo DESTINATION ./)
|
||||||
|
|
||||||
|
install(DIRECTORY models DESTINATION ./)
|
||||||
|
install(DIRECTORY images DESTINATION ./)
|
||||||
|
# install(DIRECTORY run_with_adb.sh DESTINATION ./)
|
||||||
|
|
||||||
|
file(GLOB FASTDEPLOY_LIBS ${FASTDEPLOY_INSTALL_DIR}/lib/*)
|
||||||
|
install(PROGRAMS ${FASTDEPLOY_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB OPENCV_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/opencv/lib/lib*)
|
||||||
|
install(PROGRAMS ${OPENCV_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB PADDLELITE_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/paddlelite/lib/lib*)
|
||||||
|
install(PROGRAMS ${PADDLELITE_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB TIMVX_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/paddlelite/lib/verisilicon_timvx/*)
|
||||||
|
install(PROGRAMS ${TIMVX_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB ADB_TOOLS run_with_adb.sh)
|
||||||
|
install(PROGRAMS ${ADB_TOOLS} DESTINATION ./)
|
||||||
55
examples/vision/detection/paddledetection/a311d/cpp/README.md
Executable file
55
examples/vision/detection/paddledetection/a311d/cpp/README.md
Executable file
@@ -0,0 +1,55 @@
|
|||||||
|
# PP-YOLOE 量化模型 C++ 部署示例
|
||||||
|
|
||||||
|
本目录下提供的 `infer.cc`，可以帮助用户快速完成 PP-YOLOE 量化模型在 A311D 上的部署推理加速。
|
||||||
|
|
||||||
|
## 部署准备
|
||||||
|
### FastDeploy 交叉编译环境准备
|
||||||
|
- 1. 软硬件环境满足要求，以及交叉编译环境的准备，请参考：[FastDeploy 交叉编译环境准备](../../../../../../docs/cn/build_and_install/a311d.md#交叉编译环境搭建)
|
||||||
|
|
||||||
|
### 模型准备
|
||||||
|
- 1. 用户可以直接使用由 FastDeploy 提供的量化模型进行部署。
|
||||||
|
- 2. 用户可以先使用 PaddleDetection 自行导出 Float32 模型，注意导出模型模型时设置参数：use_shared_conv=False，更多细节请参考：[PP-YOLOE](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.4/configs/ppyoloe)
|
||||||
|
- 3. 用户可以使用 FastDeploy 提供的[一键模型自动化压缩工具](../../../../../../tools/common_tools/auto_compression/)，自行进行模型量化, 并使用产出的量化模型进行部署。（注意: 推理量化后的检测模型仍然需要FP32模型文件夹下的 infer_cfg.yml 文件，自行量化的模型文件夹内不包含此 yaml 文件，用户从 FP32 模型文件夹下复制此yaml文件到量化后的模型文件夹内即可。）
|
||||||
|
- 更多量化相关相关信息可查阅[模型量化](../../quantize/README.md)
|
||||||
|
|
||||||
|
## 在 A311D 上部署量化后的 PP-YOLOE 检测模型
|
||||||
|
请按照以下步骤完成在 A311D 上部署 PP-YOLOE 量化模型：
|
||||||
|
1. 交叉编译编译 FastDeploy 库，具体请参考：[交叉编译 FastDeploy](../../../../../../docs/cn/build_and_install/a311d.md#基于-paddlelite-的-fastdeploy-交叉编译库编译)
|
||||||
|
|
||||||
|
2. 将编译后的库拷贝到当前目录，可使用如下命令：
|
||||||
|
```bash
|
||||||
|
cp -r FastDeploy/build/fastdeploy-tmivx/ FastDeploy/examples/vision/detection/yolov5/a311d/cpp
|
||||||
|
```
|
||||||
|
|
||||||
|
3. 在当前路径下载部署所需的模型和示例图片：
|
||||||
|
```bash
|
||||||
|
mkdir models && mkdir images
|
||||||
|
wget https://bj.bcebos.com/fastdeploy/models/ppyoloe_noshare_qat.tar.gz
|
||||||
|
tar -xvf ppyoloe_noshare_qat.tar.gz
|
||||||
|
cp -r ppyoloe_noshare_qat models
|
||||||
|
wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/000000014439.jpg
|
||||||
|
cp -r 000000014439.jpg images
|
||||||
|
```
|
||||||
|
|
||||||
|
4. 编译部署示例，可使入如下命令：
|
||||||
|
```bash
|
||||||
|
mkdir build && cd build
|
||||||
|
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=arm64 ..
|
||||||
|
make -j8
|
||||||
|
make install
|
||||||
|
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
||||||
|
```
|
||||||
|
|
||||||
|
5. 基于 adb 工具部署 PP-YOLOE 检测模型到晶晨 A311D
|
||||||
|
```bash
|
||||||
|
# 进入 install 目录
|
||||||
|
cd FastDeploy/examples/vision/detection/paddledetection/a311d/cpp/build/install/
|
||||||
|
# 如下命令表示：bash run_with_adb.sh 需要运行的demo 模型路径 图片路径 设备的DEVICE_ID
|
||||||
|
bash run_with_adb.sh infer_demo ppyoloe_noshare_qat 000000014439.jpg $DEVICE_ID
|
||||||
|
```
|
||||||
|
|
||||||
|
部署成功后运行结果如下：
|
||||||
|
|
||||||
|
<img width="640" src="https://user-images.githubusercontent.com/30516196/203708564-43c49485-9b48-4eb2-8fe7-0fa517979fff.png">
|
||||||
|
|
||||||
|
需要特别注意的是，在 A311D 上部署的模型需要是量化后的模型，模型的量化请参考：[模型量化](../../../../../../docs/cn/quantize.md)
|
||||||
65
examples/vision/detection/paddledetection/a311d/cpp/infer_ppyoloe.cc
Executable file
65
examples/vision/detection/paddledetection/a311d/cpp/infer_ppyoloe.cc
Executable file
@@ -0,0 +1,65 @@
|
|||||||
|
// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
|
||||||
|
//
|
||||||
|
// 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.
|
||||||
|
|
||||||
|
#include "fastdeploy/vision.h"
|
||||||
|
#ifdef WIN32
|
||||||
|
const char sep = '\\';
|
||||||
|
#else
|
||||||
|
const char sep = '/';
|
||||||
|
#endif
|
||||||
|
|
||||||
|
void InitAndInfer(const std::string& model_dir, const std::string& image_file) {
|
||||||
|
auto model_file = model_dir + sep + "model.pdmodel";
|
||||||
|
auto params_file = model_dir + sep + "model.pdiparams";
|
||||||
|
auto config_file = model_dir + sep + "infer_cfg.yml";
|
||||||
|
auto subgraph_file = model_dir + sep + "subgraph.txt";
|
||||||
|
|
||||||
|
fastdeploy::RuntimeOption option;
|
||||||
|
option.UseTimVX();
|
||||||
|
option.SetLiteSubgraphPartitionPath(subgraph_file);
|
||||||
|
|
||||||
|
auto model = fastdeploy::vision::detection::PPYOLOE(model_file, params_file,
|
||||||
|
config_file, option);
|
||||||
|
assert(model.Initialized());
|
||||||
|
|
||||||
|
auto im = cv::imread(image_file);
|
||||||
|
|
||||||
|
fastdeploy::vision::DetectionResult res;
|
||||||
|
if (!model.Predict(im, &res)) {
|
||||||
|
std::cerr << "Failed to predict." << std::endl;
|
||||||
|
return;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::cout << res.Str() << std::endl;
|
||||||
|
|
||||||
|
auto vis_im = fastdeploy::vision::VisDetection(im, res, 0.5);
|
||||||
|
cv::imwrite("vis_result.jpg", vis_im);
|
||||||
|
std::cout << "Visualized result saved in ./vis_result.jpg" << std::endl;
|
||||||
|
|
||||||
|
}
|
||||||
|
|
||||||
|
int main(int argc, char* argv[]) {
|
||||||
|
if (argc < 3) {
|
||||||
|
std::cout << "Usage: infer_demo path/to/quant_model "
|
||||||
|
"path/to/image "
|
||||||
|
"e.g ./infer_demo ./PPYOLOE_L_quant ./test.jpeg"
|
||||||
|
<< std::endl;
|
||||||
|
return -1;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::string model_dir = argv[1];
|
||||||
|
std::string test_image = argv[2];
|
||||||
|
InitAndInfer(model_dir, test_image);
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
47
examples/vision/detection/paddledetection/a311d/cpp/run_with_adb.sh
Executable file
47
examples/vision/detection/paddledetection/a311d/cpp/run_with_adb.sh
Executable file
@@ -0,0 +1,47 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
HOST_SPACE=${PWD}
|
||||||
|
echo ${HOST_SPACE}
|
||||||
|
WORK_SPACE=/data/local/tmp/test
|
||||||
|
|
||||||
|
# The first parameter represents the demo name
|
||||||
|
DEMO_NAME=image_classification_demo
|
||||||
|
if [ -n "$1" ]; then
|
||||||
|
DEMO_NAME=$1
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The second parameter represents the model name
|
||||||
|
MODEL_NAME=mobilenet_v1_fp32_224
|
||||||
|
if [ -n "$2" ]; then
|
||||||
|
MODEL_NAME=$2
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The third parameter indicates the name of the image to be tested
|
||||||
|
IMAGE_NAME=0001.jpg
|
||||||
|
if [ -n "$3" ]; then
|
||||||
|
IMAGE_NAME=$3
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The fourth parameter represents the ID of the device
|
||||||
|
ADB_DEVICE_NAME=
|
||||||
|
if [ -n "$4" ]; then
|
||||||
|
ADB_DEVICE_NAME="-s $4"
|
||||||
|
fi
|
||||||
|
|
||||||
|
# Set the environment variables required during the running process
|
||||||
|
EXPORT_ENVIRONMENT_VARIABLES="export GLOG_v=5; export SUBGRAPH_ONLINE_MODE=true; export RKNPU_LOGLEVEL=5; export RKNN_LOG_LEVEL=5; ulimit -c unlimited; export VIV_VX_ENABLE_GRAPH_TRANSFORM=-pcq:1; export VIV_VX_SET_PER_CHANNEL_ENTROPY=100; export TIMVX_BATCHNORM_FUSION_MAX_ALLOWED_QUANT_SCALE_DEVIATION=300000; export VSI_NN_LOG_LEVEL=5;"
|
||||||
|
|
||||||
|
EXPORT_ENVIRONMENT_VARIABLES="${EXPORT_ENVIRONMENT_VARIABLES}export LD_LIBRARY_PATH=${WORK_SPACE}/lib:\$LD_LIBRARY_PATH;"
|
||||||
|
|
||||||
|
# Please install adb, and DON'T run this in the docker.
|
||||||
|
set -e
|
||||||
|
adb $ADB_DEVICE_NAME shell "rm -rf $WORK_SPACE"
|
||||||
|
adb $ADB_DEVICE_NAME shell "mkdir -p $WORK_SPACE"
|
||||||
|
|
||||||
|
# Upload the demo, librarys, model and test images to the device
|
||||||
|
adb $ADB_DEVICE_NAME push ${HOST_SPACE}/lib $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push ${HOST_SPACE}/${DEMO_NAME} $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push models $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push images $WORK_SPACE
|
||||||
|
|
||||||
|
# Execute the deployment demo
|
||||||
|
adb $ADB_DEVICE_NAME shell "cd $WORK_SPACE; ${EXPORT_ENVIRONMENT_VARIABLES} chmod +x ./${DEMO_NAME}; ./${DEMO_NAME} ./models/${MODEL_NAME} ./images/$IMAGE_NAME"
|
||||||
@@ -34,7 +34,7 @@ cp -r 000000014439.jpg images
|
|||||||
4. 编译部署示例，可使入如下命令：
|
4. 编译部署示例，可使入如下命令：
|
||||||
```bash
|
```bash
|
||||||
mkdir build && cd build
|
mkdir build && cd build
|
||||||
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/timvx.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx ..
|
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=armhf ..
|
||||||
make -j8
|
make -j8
|
||||||
make install
|
make install
|
||||||
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
||||||
|
|||||||
@@ -53,7 +53,6 @@ int main(int argc, char* argv[]) {
|
|||||||
if (argc < 3) {
|
if (argc < 3) {
|
||||||
std::cout << "Usage: infer_demo path/to/quant_model "
|
std::cout << "Usage: infer_demo path/to/quant_model "
|
||||||
"path/to/image "
|
"path/to/image "
|
||||||
"run_option, "
|
|
||||||
"e.g ./infer_demo ./PPYOLOE_L_quant ./test.jpeg"
|
"e.g ./infer_demo ./PPYOLOE_L_quant ./test.jpeg"
|
||||||
<< std::endl;
|
<< std::endl;
|
||||||
return -1;
|
return -1;
|
||||||
|
|||||||
11
examples/vision/detection/yolov5/a311d/README.md
Executable file
11
examples/vision/detection/yolov5/a311d/README.md
Executable file
@@ -0,0 +1,11 @@
|
|||||||
|
# YOLOv5 量化模型在 A311D 上的部署
|
||||||
|
目前 FastDeploy 已经支持基于 PaddleLite 部署 YOLOv5 量化模型到 A311D 上。
|
||||||
|
|
||||||
|
模型的量化和量化模型的下载请参考：[模型量化](../quantize/README.md)
|
||||||
|
|
||||||
|
|
||||||
|
## 详细部署文档
|
||||||
|
|
||||||
|
在 A311D 上只支持 C++ 的部署。
|
||||||
|
|
||||||
|
- [C++部署](cpp)
|
||||||
37
examples/vision/detection/yolov5/a311d/cpp/CMakeLists.txt
Executable file
37
examples/vision/detection/yolov5/a311d/cpp/CMakeLists.txt
Executable file
@@ -0,0 +1,37 @@
|
|||||||
|
PROJECT(infer_demo C CXX)
|
||||||
|
CMAKE_MINIMUM_REQUIRED (VERSION 3.10)
|
||||||
|
|
||||||
|
# 指定下载解压后的fastdeploy库路径
|
||||||
|
option(FASTDEPLOY_INSTALL_DIR "Path of downloaded fastdeploy sdk.")
|
||||||
|
|
||||||
|
include(${FASTDEPLOY_INSTALL_DIR}/FastDeploy.cmake)
|
||||||
|
|
||||||
|
# 添加FastDeploy依赖头文件
|
||||||
|
include_directories(${FASTDEPLOY_INCS})
|
||||||
|
include_directories(${FastDeploy_INCLUDE_DIRS})
|
||||||
|
|
||||||
|
add_executable(infer_demo ${PROJECT_SOURCE_DIR}/infer.cc)
|
||||||
|
# 添加FastDeploy库依赖
|
||||||
|
target_link_libraries(infer_demo ${FASTDEPLOY_LIBS})
|
||||||
|
|
||||||
|
set(CMAKE_INSTALL_PREFIX ${CMAKE_SOURCE_DIR}/build/install)
|
||||||
|
|
||||||
|
install(TARGETS infer_demo DESTINATION ./)
|
||||||
|
|
||||||
|
install(DIRECTORY models DESTINATION ./)
|
||||||
|
install(DIRECTORY images DESTINATION ./)
|
||||||
|
|
||||||
|
file(GLOB FASTDEPLOY_LIBS ${FASTDEPLOY_INSTALL_DIR}/lib/*)
|
||||||
|
install(PROGRAMS ${FASTDEPLOY_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB OPENCV_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/opencv/lib/lib*)
|
||||||
|
install(PROGRAMS ${OPENCV_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB PADDLELITE_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/paddlelite/lib/lib*)
|
||||||
|
install(PROGRAMS ${PADDLELITE_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB TIMVX_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/paddlelite/lib/verisilicon_timvx/*)
|
||||||
|
install(PROGRAMS ${TIMVX_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB ADB_TOOLS run_with_adb.sh)
|
||||||
|
install(PROGRAMS ${ADB_TOOLS} DESTINATION ./)
|
||||||
54
examples/vision/detection/yolov5/a311d/cpp/README.md
Executable file
54
examples/vision/detection/yolov5/a311d/cpp/README.md
Executable file
@@ -0,0 +1,54 @@
|
|||||||
|
# YOLOv5 量化模型 C++ 部署示例
|
||||||
|
|
||||||
|
本目录下提供的 `infer.cc`，可以帮助用户快速完成 YOLOv5 量化模型在 A311D 上的部署推理加速。
|
||||||
|
|
||||||
|
## 部署准备
|
||||||
|
### FastDeploy 交叉编译环境准备
|
||||||
|
- 1. 软硬件环境满足要求，以及交叉编译环境的准备，请参考：[FastDeploy 交叉编译环境准备](../../../../../../docs/cn/build_and_install/a311d.md#交叉编译环境搭建)
|
||||||
|
|
||||||
|
### 量化模型准备
|
||||||
|
- 1. 用户可以直接使用由 FastDeploy 提供的量化模型进行部署。
|
||||||
|
- 2. 用户可以使用 FastDeploy 提供的[一键模型自动化压缩工具](../../../../../../tools/common_tools/auto_compression/),自行进行模型量化, 并使用产出的量化模型进行部署。
|
||||||
|
- 更多量化相关相关信息可查阅[模型量化](../../quantize/README.md)
|
||||||
|
|
||||||
|
## 在 A311D 上部署量化后的 YOLOv5 检测模型
|
||||||
|
请按照以下步骤完成在 A311D 上部署 YOLOv5 量化模型：
|
||||||
|
1. 交叉编译编译 FastDeploy 库，具体请参考：[交叉编译 FastDeploy](../../../../../../docs/cn/build_and_install/a311d.md#基于-paddlelite-的-fastdeploy-交叉编译库编译)
|
||||||
|
|
||||||
|
2. 将编译后的库拷贝到当前目录，可使用如下命令：
|
||||||
|
```bash
|
||||||
|
cp -r FastDeploy/build/fastdeploy-tmivx/ FastDeploy/examples/vision/detection/yolov5/a311d/cpp
|
||||||
|
```
|
||||||
|
|
||||||
|
3. 在当前路径下载部署所需的模型和示例图片：
|
||||||
|
```bash
|
||||||
|
mkdir models && mkdir images
|
||||||
|
wget https://bj.bcebos.com/fastdeploy/models/yolov5s_ptq_model.tar.gz
|
||||||
|
tar -xvf yolov5s_ptq_model.tar.gz
|
||||||
|
cp -r yolov5s_ptq_model models
|
||||||
|
wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/000000014439.jpg
|
||||||
|
cp -r 000000014439.jpg images
|
||||||
|
```
|
||||||
|
|
||||||
|
4. 编译部署示例，可使入如下命令：
|
||||||
|
```bash
|
||||||
|
mkdir build && cd build
|
||||||
|
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=arm64 ..
|
||||||
|
make -j8
|
||||||
|
make install
|
||||||
|
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
||||||
|
```
|
||||||
|
|
||||||
|
5. 基于 adb 工具部署 YOLOv5 检测模型到晶晨 A311D
|
||||||
|
```bash
|
||||||
|
# 进入 install 目录
|
||||||
|
cd FastDeploy/examples/vision/detection/yolov5/a311d/cpp/build/install/
|
||||||
|
# 如下命令表示：bash run_with_adb.sh 需要运行的demo 模型路径 图片路径 设备的DEVICE_ID
|
||||||
|
bash run_with_adb.sh infer_demo yolov5s_ptq_model 000000014439.jpg $DEVICE_ID
|
||||||
|
```
|
||||||
|
|
||||||
|
部署成功后，vis_result.jpg 保存的结果如下：
|
||||||
|
|
||||||
|
<img width="640" src="https://user-images.githubusercontent.com/30516196/203706969-dd58493c-6635-4ee7-9421-41c2e0c9524b.png">
|
||||||
|
|
||||||
|
需要特别注意的是，在 A311D 上部署的模型需要是量化后的模型，模型的量化请参考：[模型量化](../../../../../../docs/cn/quantize.md)
|
||||||
64
examples/vision/detection/yolov5/a311d/cpp/infer.cc
Executable file
64
examples/vision/detection/yolov5/a311d/cpp/infer.cc
Executable file
@@ -0,0 +1,64 @@
|
|||||||
|
// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
|
||||||
|
//
|
||||||
|
// 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.
|
||||||
|
|
||||||
|
#include "fastdeploy/vision.h"
|
||||||
|
#ifdef WIN32
|
||||||
|
const char sep = '\\';
|
||||||
|
#else
|
||||||
|
const char sep = '/';
|
||||||
|
#endif
|
||||||
|
|
||||||
|
void InitAndInfer(const std::string& model_dir, const std::string& image_file) {
|
||||||
|
auto model_file = model_dir + sep + "model.pdmodel";
|
||||||
|
auto params_file = model_dir + sep + "model.pdiparams";
|
||||||
|
auto subgraph_file = model_dir + sep + "subgraph.txt";
|
||||||
|
|
||||||
|
fastdeploy::RuntimeOption option;
|
||||||
|
option.UseTimVX();
|
||||||
|
option.SetLiteSubgraphPartitionPath(subgraph_file);
|
||||||
|
|
||||||
|
auto model = fastdeploy::vision::detection::YOLOv5(
|
||||||
|
model_file, params_file, option, fastdeploy::ModelFormat::PADDLE);
|
||||||
|
assert(model.Initialized());
|
||||||
|
|
||||||
|
auto im = cv::imread(image_file);
|
||||||
|
|
||||||
|
fastdeploy::vision::DetectionResult res;
|
||||||
|
if (!model.Predict(im, &res)) {
|
||||||
|
std::cerr << "Failed to predict." << std::endl;
|
||||||
|
return;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::cout << res.Str() << std::endl;
|
||||||
|
|
||||||
|
auto vis_im = fastdeploy::vision::VisDetection(im, res);
|
||||||
|
cv::imwrite("vis_result.jpg", vis_im);
|
||||||
|
std::cout << "Visualized result saved in ./vis_result.jpg" << std::endl;
|
||||||
|
}
|
||||||
|
|
||||||
|
int main(int argc, char* argv[]) {
|
||||||
|
if (argc < 3) {
|
||||||
|
std::cout << "Usage: infer_demo path/to/quant_model "
|
||||||
|
"path/to/image "
|
||||||
|
"run_option, "
|
||||||
|
"e.g ./infer_demo ./yolov5s_quant ./000000014439.jpg"
|
||||||
|
<< std::endl;
|
||||||
|
return -1;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::string model_dir = argv[1];
|
||||||
|
std::string test_image = argv[2];
|
||||||
|
InitAndInfer(model_dir, test_image);
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
47
examples/vision/detection/yolov5/a311d/cpp/run_with_adb.sh
Executable file
47
examples/vision/detection/yolov5/a311d/cpp/run_with_adb.sh
Executable file
@@ -0,0 +1,47 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
HOST_SPACE=${PWD}
|
||||||
|
echo ${HOST_SPACE}
|
||||||
|
WORK_SPACE=/data/local/tmp/test
|
||||||
|
|
||||||
|
# The first parameter represents the demo name
|
||||||
|
DEMO_NAME=image_classification_demo
|
||||||
|
if [ -n "$1" ]; then
|
||||||
|
DEMO_NAME=$1
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The second parameter represents the model name
|
||||||
|
MODEL_NAME=mobilenet_v1_fp32_224
|
||||||
|
if [ -n "$2" ]; then
|
||||||
|
MODEL_NAME=$2
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The third parameter indicates the name of the image to be tested
|
||||||
|
IMAGE_NAME=0001.jpg
|
||||||
|
if [ -n "$3" ]; then
|
||||||
|
IMAGE_NAME=$3
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The fourth parameter represents the ID of the device
|
||||||
|
ADB_DEVICE_NAME=
|
||||||
|
if [ -n "$4" ]; then
|
||||||
|
ADB_DEVICE_NAME="-s $4"
|
||||||
|
fi
|
||||||
|
|
||||||
|
# Set the environment variables required during the running process
|
||||||
|
EXPORT_ENVIRONMENT_VARIABLES="export GLOG_v=5; export VIV_VX_ENABLE_GRAPH_TRANSFORM=-pcq:1; export VIV_VX_SET_PER_CHANNEL_ENTROPY=100; export TIMVX_BATCHNORM_FUSION_MAX_ALLOWED_QUANT_SCALE_DEVIATION=300000; export VSI_NN_LOG_LEVEL=5;"
|
||||||
|
|
||||||
|
EXPORT_ENVIRONMENT_VARIABLES="${EXPORT_ENVIRONMENT_VARIABLES}export LD_LIBRARY_PATH=${WORK_SPACE}/lib:\$LD_LIBRARY_PATH;"
|
||||||
|
|
||||||
|
# Please install adb, and DON'T run this in the docker.
|
||||||
|
set -e
|
||||||
|
adb $ADB_DEVICE_NAME shell "rm -rf $WORK_SPACE"
|
||||||
|
adb $ADB_DEVICE_NAME shell "mkdir -p $WORK_SPACE"
|
||||||
|
|
||||||
|
# Upload the demo, librarys, model and test images to the device
|
||||||
|
adb $ADB_DEVICE_NAME push ${HOST_SPACE}/lib $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push ${HOST_SPACE}/${DEMO_NAME} $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push models $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push images $WORK_SPACE
|
||||||
|
|
||||||
|
# Execute the deployment demo
|
||||||
|
adb $ADB_DEVICE_NAME shell "cd $WORK_SPACE; ${EXPORT_ENVIRONMENT_VARIABLES} chmod +x ./${DEMO_NAME}; ./${DEMO_NAME} ./models/${MODEL_NAME} ./images/$IMAGE_NAME"
|
||||||
@@ -33,7 +33,7 @@ cp -r 000000014439.jpg images
|
|||||||
4. 编译部署示例，可使入如下命令：
|
4. 编译部署示例，可使入如下命令：
|
||||||
```bash
|
```bash
|
||||||
mkdir build && cd build
|
mkdir build && cd build
|
||||||
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/timvx.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx ..
|
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=armhf ..
|
||||||
make -j8
|
make -j8
|
||||||
make install
|
make install
|
||||||
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
||||||
|
|||||||
11
examples/vision/segmentation/paddleseg/a311d/README.md
Executable file
11
examples/vision/segmentation/paddleseg/a311d/README.md
Executable file
@@ -0,0 +1,11 @@
|
|||||||
|
# PP-LiteSeg 量化模型在 A311D 上的部署
|
||||||
|
目前 FastDeploy 已经支持基于 PaddleLite 部署 PP-LiteSeg 量化模型到 A311D 上。
|
||||||
|
|
||||||
|
模型的量化和量化模型的下载请参考：[模型量化](../quantize/README.md)
|
||||||
|
|
||||||
|
|
||||||
|
## 详细部署文档
|
||||||
|
|
||||||
|
在 A311D 上只支持 C++ 的部署。
|
||||||
|
|
||||||
|
- [C++部署](cpp)
|
||||||
38
examples/vision/segmentation/paddleseg/a311d/cpp/CMakeLists.txt
Executable file
38
examples/vision/segmentation/paddleseg/a311d/cpp/CMakeLists.txt
Executable file
@@ -0,0 +1,38 @@
|
|||||||
|
PROJECT(infer_demo C CXX)
|
||||||
|
CMAKE_MINIMUM_REQUIRED (VERSION 3.10)
|
||||||
|
|
||||||
|
# 指定下载解压后的fastdeploy库路径
|
||||||
|
option(FASTDEPLOY_INSTALL_DIR "Path of downloaded fastdeploy sdk.")
|
||||||
|
|
||||||
|
include(${FASTDEPLOY_INSTALL_DIR}/FastDeploy.cmake)
|
||||||
|
|
||||||
|
# 添加FastDeploy依赖头文件
|
||||||
|
include_directories(${FASTDEPLOY_INCS})
|
||||||
|
include_directories(${FastDeploy_INCLUDE_DIRS})
|
||||||
|
|
||||||
|
add_executable(infer_demo ${PROJECT_SOURCE_DIR}/infer.cc)
|
||||||
|
# 添加FastDeploy库依赖
|
||||||
|
target_link_libraries(infer_demo ${FASTDEPLOY_LIBS})
|
||||||
|
|
||||||
|
set(CMAKE_INSTALL_PREFIX ${CMAKE_SOURCE_DIR}/build/install)
|
||||||
|
|
||||||
|
install(TARGETS infer_demo DESTINATION ./)
|
||||||
|
|
||||||
|
install(DIRECTORY models DESTINATION ./)
|
||||||
|
install(DIRECTORY images DESTINATION ./)
|
||||||
|
# install(DIRECTORY run_with_adb.sh DESTINATION ./)
|
||||||
|
|
||||||
|
file(GLOB FASTDEPLOY_LIBS ${FASTDEPLOY_INSTALL_DIR}/lib/*)
|
||||||
|
install(PROGRAMS ${FASTDEPLOY_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB OPENCV_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/opencv/lib/lib*)
|
||||||
|
install(PROGRAMS ${OPENCV_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB PADDLELITE_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/paddlelite/lib/lib*)
|
||||||
|
install(PROGRAMS ${PADDLELITE_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB TIMVX_LIBS ${FASTDEPLOY_INSTALL_DIR}/third_libs/install/paddlelite/lib/verisilicon_timvx/*)
|
||||||
|
install(PROGRAMS ${TIMVX_LIBS} DESTINATION lib)
|
||||||
|
|
||||||
|
file(GLOB ADB_TOOLS run_with_adb.sh)
|
||||||
|
install(PROGRAMS ${ADB_TOOLS} DESTINATION ./)
|
||||||
54
examples/vision/segmentation/paddleseg/a311d/cpp/README.md
Executable file
54
examples/vision/segmentation/paddleseg/a311d/cpp/README.md
Executable file
@@ -0,0 +1,54 @@
|
|||||||
|
# PP-LiteSeg 量化模型 C++ 部署示例
|
||||||
|
|
||||||
|
本目录下提供的 `infer.cc`，可以帮助用户快速完成 PP-LiteSeg 量化模型在 A311D 上的部署推理加速。
|
||||||
|
|
||||||
|
## 部署准备
|
||||||
|
### FastDeploy 交叉编译环境准备
|
||||||
|
- 1. 软硬件环境满足要求，以及交叉编译环境的准备，请参考：[FastDeploy 交叉编译环境准备](../../../../../../docs/cn/build_and_install/a311d.md#交叉编译环境搭建)
|
||||||
|
|
||||||
|
### 模型准备
|
||||||
|
- 1. 用户可以直接使用由 FastDeploy 提供的量化模型进行部署。
|
||||||
|
- 2. 用户可以使用 FastDeploy 提供的一键模型自动化压缩工具,自行进行模型量化, 并使用产出的量化模型进行部署.(注意: 推理量化后的分类模型仍然需要FP32模型文件夹下的 deploy.yaml 文件, 自行量化的模型文件夹内不包含此 yaml 文件, 用户从FP32模型文件夹下复制此yaml文件到量化后的模型文件夹内即可.)
|
||||||
|
- 更多量化相关相关信息可查阅[模型量化](../../quantize/README.md)
|
||||||
|
|
||||||
|
## 在 A311D 上部署量化后的 PP-LiteSeg 分割模型
|
||||||
|
请按照以下步骤完成在 A311D 上部署 PP-LiteSeg 量化模型：
|
||||||
|
1. 交叉编译编译 FastDeploy 库，具体请参考：[交叉编译 FastDeploy](../../../../../../docs/cn/build_and_install/a311d.md#基于-paddlelite-的-fastdeploy-交叉编译库编译)
|
||||||
|
|
||||||
|
2. 将编译后的库拷贝到当前目录，可使用如下命令：
|
||||||
|
```bash
|
||||||
|
cp -r FastDeploy/build/fastdeploy-tmivx/ FastDeploy/examples/vision/segmentation/paddleseg/a311d/cpp
|
||||||
|
```
|
||||||
|
|
||||||
|
3. 在当前路径下载部署所需的模型和示例图片：
|
||||||
|
```bash
|
||||||
|
mkdir models && mkdir images
|
||||||
|
wget https://bj.bcebos.com/fastdeploy/models/rk1/ppliteseg.tar.gz
|
||||||
|
tar -xvf ppliteseg.tar.gz
|
||||||
|
cp -r ppliteseg models
|
||||||
|
wget https://paddleseg.bj.bcebos.com/dygraph/demo/cityscapes_demo.png
|
||||||
|
cp -r cityscapes_demo.png images
|
||||||
|
```
|
||||||
|
|
||||||
|
4. 编译部署示例，可使入如下命令：
|
||||||
|
```bash
|
||||||
|
mkdir build && cd build
|
||||||
|
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=arm64 ..
|
||||||
|
make -j8
|
||||||
|
make install
|
||||||
|
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
||||||
|
```
|
||||||
|
|
||||||
|
5. 基于 adb 工具部署 PP-LiteSeg 分割模型到晶晨 A311D，可使用如下命令：
|
||||||
|
```bash
|
||||||
|
# 进入 install 目录
|
||||||
|
cd FastDeploy/examples/vision/segmentation/paddleseg/a311d/cpp/build/install/
|
||||||
|
# 如下命令表示：bash run_with_adb.sh 需要运行的demo 模型路径 图片路径 设备的DEVICE_ID
|
||||||
|
bash run_with_adb.sh infer_demo ppliteseg cityscapes_demo.png $DEVICE_ID
|
||||||
|
```
|
||||||
|
|
||||||
|
部署成功后运行结果如下：
|
||||||
|
|
||||||
|
<img width="640" src="https://user-images.githubusercontent.com/30516196/205544166-9b2719ff-ed82-4908-b90a-095de47392e1.png">
|
||||||
|
|
||||||
|
需要特别注意的是，在 A311D 上部署的模型需要是量化后的模型，模型的量化请参考：[模型量化](../../../../../../docs/cn/quantize.md)
|
||||||
65
examples/vision/segmentation/paddleseg/a311d/cpp/infer.cc
Executable file
65
examples/vision/segmentation/paddleseg/a311d/cpp/infer.cc
Executable file
@@ -0,0 +1,65 @@
|
|||||||
|
// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
|
||||||
|
//
|
||||||
|
// 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.
|
||||||
|
|
||||||
|
#include "fastdeploy/vision.h"
|
||||||
|
#ifdef WIN32
|
||||||
|
const char sep = '\\';
|
||||||
|
#else
|
||||||
|
const char sep = '/';
|
||||||
|
#endif
|
||||||
|
|
||||||
|
void InitAndInfer(const std::string& model_dir, const std::string& image_file) {
|
||||||
|
auto model_file = model_dir + sep + "model.pdmodel";
|
||||||
|
auto params_file = model_dir + sep + "model.pdiparams";
|
||||||
|
auto config_file = model_dir + sep + "deploy.yaml";
|
||||||
|
auto subgraph_file = model_dir + sep + "subgraph.txt";
|
||||||
|
|
||||||
|
fastdeploy::RuntimeOption option;
|
||||||
|
option.UseTimVX();
|
||||||
|
option.SetLiteSubgraphPartitionPath(subgraph_file);
|
||||||
|
|
||||||
|
auto model = fastdeploy::vision::segmentation::PaddleSegModel(
|
||||||
|
model_file, params_file, config_file,option);
|
||||||
|
|
||||||
|
assert(model.Initialized());
|
||||||
|
|
||||||
|
auto im = cv::imread(image_file);
|
||||||
|
|
||||||
|
fastdeploy::vision::SegmentationResult res;
|
||||||
|
if (!model.Predict(im, &res)) {
|
||||||
|
std::cerr << "Failed to predict." << std::endl;
|
||||||
|
return;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::cout << res.Str() << std::endl;
|
||||||
|
|
||||||
|
auto vis_im = fastdeploy::vision::VisSegmentation(im, res, 0.5);
|
||||||
|
cv::imwrite("vis_result.jpg", vis_im);
|
||||||
|
std::cout << "Visualized result saved in ./vis_result.jpg" << std::endl;
|
||||||
|
}
|
||||||
|
|
||||||
|
int main(int argc, char* argv[]) {
|
||||||
|
if (argc < 3) {
|
||||||
|
std::cout << "Usage: infer_demo path/to/quant_model "
|
||||||
|
"path/to/image "
|
||||||
|
"e.g ./infer_demo ./ResNet50_vd_quant ./test.jpeg"
|
||||||
|
<< std::endl;
|
||||||
|
return -1;
|
||||||
|
}
|
||||||
|
|
||||||
|
std::string model_dir = argv[1];
|
||||||
|
std::string test_image = argv[2];
|
||||||
|
InitAndInfer(model_dir, test_image);
|
||||||
|
return 0;
|
||||||
|
}
|
||||||
47
examples/vision/segmentation/paddleseg/a311d/cpp/run_with_adb.sh
Executable file
47
examples/vision/segmentation/paddleseg/a311d/cpp/run_with_adb.sh
Executable file
@@ -0,0 +1,47 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
HOST_SPACE=${PWD}
|
||||||
|
echo ${HOST_SPACE}
|
||||||
|
WORK_SPACE=/data/local/tmp/test
|
||||||
|
|
||||||
|
# The first parameter represents the demo name
|
||||||
|
DEMO_NAME=image_classification_demo
|
||||||
|
if [ -n "$1" ]; then
|
||||||
|
DEMO_NAME=$1
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The second parameter represents the model name
|
||||||
|
MODEL_NAME=mobilenet_v1_fp32_224
|
||||||
|
if [ -n "$2" ]; then
|
||||||
|
MODEL_NAME=$2
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The third parameter indicates the name of the image to be tested
|
||||||
|
IMAGE_NAME=0001.jpg
|
||||||
|
if [ -n "$3" ]; then
|
||||||
|
IMAGE_NAME=$3
|
||||||
|
fi
|
||||||
|
|
||||||
|
# The fourth parameter represents the ID of the device
|
||||||
|
ADB_DEVICE_NAME=
|
||||||
|
if [ -n "$4" ]; then
|
||||||
|
ADB_DEVICE_NAME="-s $4"
|
||||||
|
fi
|
||||||
|
|
||||||
|
# Set the environment variables required during the running process
|
||||||
|
EXPORT_ENVIRONMENT_VARIABLES="export GLOG_v=5; export VIV_VX_ENABLE_GRAPH_TRANSFORM=-pcq:1; export VIV_VX_SET_PER_CHANNEL_ENTROPY=100; export TIMVX_BATCHNORM_FUSION_MAX_ALLOWED_QUANT_SCALE_DEVIATION=300000; export VSI_NN_LOG_LEVEL=5;"
|
||||||
|
|
||||||
|
EXPORT_ENVIRONMENT_VARIABLES="${EXPORT_ENVIRONMENT_VARIABLES}export LD_LIBRARY_PATH=${WORK_SPACE}/lib:\$LD_LIBRARY_PATH;"
|
||||||
|
|
||||||
|
# Please install adb, and DON'T run this in the docker.
|
||||||
|
set -e
|
||||||
|
adb $ADB_DEVICE_NAME shell "rm -rf $WORK_SPACE"
|
||||||
|
adb $ADB_DEVICE_NAME shell "mkdir -p $WORK_SPACE"
|
||||||
|
|
||||||
|
# Upload the demo, librarys, model and test images to the device
|
||||||
|
adb $ADB_DEVICE_NAME push ${HOST_SPACE}/lib $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push ${HOST_SPACE}/${DEMO_NAME} $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push models $WORK_SPACE
|
||||||
|
adb $ADB_DEVICE_NAME push images $WORK_SPACE
|
||||||
|
|
||||||
|
# Execute the deployment demo
|
||||||
|
adb $ADB_DEVICE_NAME shell "cd $WORK_SPACE; ${EXPORT_ENVIRONMENT_VARIABLES} chmod +x ./${DEMO_NAME}; ./${DEMO_NAME} ./models/${MODEL_NAME} ./images/$IMAGE_NAME"
|
||||||
@@ -33,7 +33,7 @@ cp -r cityscapes_demo.png images
|
|||||||
4. 编译部署示例，可使入如下命令：
|
4. 编译部署示例，可使入如下命令：
|
||||||
```bash
|
```bash
|
||||||
mkdir build && cd build
|
mkdir build && cd build
|
||||||
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/timvx.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx ..
|
cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=armhf ..
|
||||||
make -j8
|
make -j8
|
||||||
make install
|
make install
|
||||||
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
# 成功编译之后，会生成 install 文件夹，里面有一个运行 demo 和部署所需的库
|
||||||
|
|||||||
@@ -53,7 +53,6 @@ int main(int argc, char* argv[]) {
|
|||||||
if (argc < 3) {
|
if (argc < 3) {
|
||||||
std::cout << "Usage: infer_demo path/to/quant_model "
|
std::cout << "Usage: infer_demo path/to/quant_model "
|
||||||
"path/to/image "
|
"path/to/image "
|
||||||
"run_option, "
|
|
||||||
"e.g ./infer_demo ./ResNet50_vd_quant ./test.jpeg"
|
"e.g ./infer_demo ./ResNet50_vd_quant ./test.jpeg"
|
||||||
<< std::endl;
|
<< std::endl;
|
||||||
return -1;
|
return -1;
|
||||||
|
|||||||
@@ -12,6 +12,7 @@
|
|||||||
// See the License for the specific language governing permissions and
|
// See the License for the specific language governing permissions and
|
||||||
// limitations under the License.
|
// limitations under the License.
|
||||||
#pragma once
|
#pragma once
|
||||||
|
#include <stdint.h>
|
||||||
#include <cmath>
|
#include <cmath>
|
||||||
#include <vector>
|
#include <vector>
|
||||||
|
|
||||||
|
|||||||
17
tools/timvx/install.sh
Normal file
17
tools/timvx/install.sh
Normal file
@@ -0,0 +1,17 @@
|
|||||||
|
# 1. Install basic software
|
||||||
|
apt update
|
||||||
|
apt-get install -y --no-install-recommends \
|
||||||
|
gcc g++ git make wget python unzip
|
||||||
|
|
||||||
|
# 2. Install arm gcc toolchains
|
||||||
|
apt-get install -y --no-install-recommends \
|
||||||
|
g++-arm-linux-gnueabi gcc-arm-linux-gnueabi \
|
||||||
|
g++-arm-linux-gnueabihf gcc-arm-linux-gnueabihf \
|
||||||
|
gcc-aarch64-linux-gnu g++-aarch64-linux-gnu
|
||||||
|
|
||||||
|
# 3. Install cmake 3.10 or above
|
||||||
|
wget -c https://mms-res.cdn.bcebos.com/cmake-3.10.3-Linux-x86_64.tar.gz && \
|
||||||
|
tar xzf cmake-3.10.3-Linux-x86_64.tar.gz && \
|
||||||
|
mv cmake-3.10.3-Linux-x86_64 /opt/cmake-3.10 && \
|
||||||
|
ln -s /opt/cmake-3.10/bin/cmake /usr/bin/cmake && \
|
||||||
|
ln -s /opt/cmake-3.10/bin/ccmake /usr/bin/ccmake
|
||||||
Reference in New Issue
Block a user