diff --git a/cmake/opencv.cmake b/cmake/opencv.cmake
index 87f8c8bcd..fd2ecabe4 100755
--- a/cmake/opencv.cmake
+++ b/cmake/opencv.cmake
@@ -41,12 +41,6 @@ elseif(IOS)
else()
if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
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()
if(ENABLE_OPENCV_CUDA)
if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
@@ -56,15 +50,20 @@ else()
endif()
endif()
+if(NOT OPENCV_FILENAME)
+ set(OPENCV_FILENAME "opencv-linux-x64-3.4.16")
+endif()
+
set(OPENCV_INSTALL_DIR ${THIRD_PARTY_PATH}/install/)
if(ANDROID)
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.
set(OPENCV_URL_PREFIX "https://bj.bcebos.com/paddle2onnx/libs")
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(EXISTS /usr/lib/aarch64-linux-gnu/cmake/opencv4/)
@@ -186,9 +185,8 @@ else()
endif()
file(RENAME ${THIRD_PARTY_PATH}/install/${OPENCV_FILENAME}/ ${THIRD_PARTY_PATH}/install/opencv)
set(OPENCV_FILENAME opencv)
- set(OpenCV_DIR ${THIRD_PARTY_PATH}/install/${OPENCV_FILENAME})
- if(ENABLE_TIMVX)
- set(OpenCV_DIR ${OpenCV_DIR}/lib/cmake/opencv4)
+ if(NOT OpenCV_DIR)
+ set(OpenCV_DIR ${THIRD_PARTY_PATH}/install/${OPENCV_FILENAME})
endif()
if (WIN32)
set(OpenCV_DIR ${OpenCV_DIR}/build)
diff --git a/cmake/paddlelite.cmake b/cmake/paddlelite.cmake
index bcc0eb470..74525b7a9 100755
--- a/cmake/paddlelite.cmake
+++ b/cmake/paddlelite.cmake
@@ -49,22 +49,20 @@ if(ANDROID)
endif()
endif()
-if(WIN32 OR APPLE OR IOS)
- message(FATAL_ERROR "Doesn't support windows/mac/ios platform with backend Paddle Lite now.")
-elseif(ANDROID)
- set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-android-${ANDROID_ABI}-latest-dev.tgz")
- if(ANDROID_ABI MATCHES "arm64-v8a")
- set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-android-${ANDROID_ABI}-fp16-latest-dev.tgz")
- endif()
-else() # Linux
- if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
- set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-linux-arm64-20221209.tgz")
- else()
- if(ENABLE_TIMVX)
- set(PADDLELITE_URL "https://bj.bcebos.com/fastdeploy/test/lite-linux_armhf_1130.tgz")
+if(NOT PADDLELITE_URL)
+ if(WIN32 OR APPLE OR IOS)
+ message(FATAL_ERROR "Doesn't support windows/mac/ios platform with backend Paddle Lite now.")
+ elseif(ANDROID)
+ set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-android-${ANDROID_ABI}-latest-dev.tgz")
+ if(ANDROID_ABI MATCHES "arm64-v8a")
+ set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-android-${ANDROID_ABI}-fp16-latest-dev.tgz")
+ endif()
+ else() # Linux
+ if(CMAKE_HOST_SYSTEM_PROCESSOR MATCHES "aarch64")
+ set(PADDLELITE_URL "${PADDLELITE_URL_PREFIX}/lite-linux-arm64-20221209.tgz")
else()
- message(FATAL_ERROR "Only support Linux aarch64 or ENABLE_TIMVX now, x64 is not supported with backend Paddle Lite.")
- endif()
+ message(FATAL_ERROR "Only support Linux aarch64 now, x64 is not supported with backend Paddle Lite.")
+ endif()
endif()
endif()
diff --git a/cmake/timvx.cmake b/cmake/timvx.cmake
index c6a7d54d2..973face96 100755
--- a/cmake/timvx.cmake
+++ b/cmake/timvx.cmake
@@ -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(${WITH_GPU})
- message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DWITH_GPU=OFF")
- set(WITH_GPU OFF)
- endif()
-
- if(${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()
- if (DEFINED CMAKE_INSTALL_PREFIX)
- install(FILES ${PROJECT_SOURCE_DIR}/cmake/timvx.cmake DESTINATION ${CMAKE_INSTALL_PREFIX})
- endif()
+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(${WITH_GPU})
+ message(WARNING "While compiling with -DWITH_TIMVX=ON, will force to set -DWITH_GPU=OFF")
+ set(WITH_GPU OFF)
+endif()
+
+if(${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 ${CMAKE_INSTALL_PREFIX})
+
diff --git a/cmake/toolchain.cmake b/cmake/toolchain.cmake
new file mode 100755
index 000000000..4b3485748
--- /dev/null
+++ b/cmake/toolchain.cmake
@@ -0,0 +1,38 @@
+if (DEFINED TARGET_ABI)
+ set(CMAKE_SYSTEM_NAME Linux)
+ set(CMAKE_BUILD_TYPE MinSizeRel)
+ if(${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()
+
diff --git a/docs/cn/build_and_install/README.md b/docs/cn/build_and_install/README.md
index 8be1f745a..7ec07c7b8 100755
--- a/docs/cn/build_and_install/README.md
+++ b/docs/cn/build_and_install/README.md
@@ -12,6 +12,7 @@
- [Jetson部署环境](jetson.md)
- [Android平台部署环境](android.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
@@ -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_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_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_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 |
diff --git a/docs/cn/build_and_install/a311d.md b/docs/cn/build_and_install/a311d.md
new file mode 100755
index 000000000..4b3773f0d
--- /dev/null
+++ b/docs/cn/build_and_install/a311d.md
@@ -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)
diff --git a/docs/cn/build_and_install/rv1126.md b/docs/cn/build_and_install/rv1126.md
index f3cd4ed6a..ff0050715 100755
--- a/docs/cn/build_and_install/rv1126.md
+++ b/docs/cn/build_and_install/rv1126.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|
|:---|:---|:---|:---|
|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)
@@ -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
### 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
@@ -49,8 +55,9 @@ cd FastDeploy
mkdir build && cd build
# CMake configuration with RK toolchain
-cmake -DCMAKE_TOOLCHAIN_FILE=./../cmake/timvx.cmake \
- -DENABLE_TIMVX=ON \
+cmake -DCMAKE_TOOLCHAIN_FILE=./../cmake/toolchain.cmake \
+ -DWITH_TIMVX=ON \
+ -DTARGET_ABI=armhf \
-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 ..
diff --git a/examples/vision/classification/paddleclas/a311d/README.md b/examples/vision/classification/paddleclas/a311d/README.md
new file mode 100755
index 000000000..0fb75854a
--- /dev/null
+++ b/examples/vision/classification/paddleclas/a311d/README.md
@@ -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)
diff --git a/examples/vision/classification/paddleclas/a311d/cpp/CMakeLists.txt b/examples/vision/classification/paddleclas/a311d/cpp/CMakeLists.txt
new file mode 100755
index 000000000..baaf8331f
--- /dev/null
+++ b/examples/vision/classification/paddleclas/a311d/cpp/CMakeLists.txt
@@ -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 ${ADB_TOOLS} DESTINATION ./)
diff --git a/examples/vision/classification/paddleclas/a311d/cpp/README.md b/examples/vision/classification/paddleclas/a311d/cpp/README.md
new file mode 100755
index 000000000..c1d926c9a
--- /dev/null
+++ b/examples/vision/classification/paddleclas/a311d/cpp/README.md
@@ -0,0 +1,53 @@
+# PaddleClas A311D 开发板 C++ 部署示例
+本目录下提供的 `infer.cc`,可以帮助用户快速完成 PaddleClas 量化模型在 A311D 上的部署推理加速。
+
+## 部署准备
+### FastDeploy 交叉编译环境准备
+- 1. 软硬件环境满足要求,以及交叉编译环境的准备,请参考:[FastDeploy 交叉编译环境准备](../../../../../../docs/cn/build_and_install/a311d.md#交叉编译环境搭建)
+
+### 量化模型准备
+- 1. 用户可以直接使用由 FastDeploy 提供的量化模型进行部署。
+- 2. 用户可以使用 FastDeploy 提供的[一键模型自动化压缩工具](../../../../../../tools/common_tools/auto_compression/),自行进行模型量化, 并使用产出的量化模型进行部署。(注意: 推理量化后的分类模型仍然需要FP32模型文件夹下的inference_cls.yaml文件, 自行量化的模型文件夹内不包含此 yaml 文件, 用户从 FP32 模型文件夹下复制此 yaml 文件到量化后的模型文件夹内即可.)
+- 更多量化相关相关信息可查阅[模型量化](../../quantize/README.md)
+
+## 在 A311D 上部署量化后的 ResNet50_Vd 分类模型
+请按照以下步骤完成在 A311D 上部署 ResNet50_Vd 量化模型:
+1. 交叉编译编译 FastDeploy 库,具体请参考:[交叉编译 FastDeploy](../../../../../../docs/cn/build_and_install/a311d.md#基于-paddlelite-的-fastdeploy-交叉编译库编译)
+
+2. 将编译后的库拷贝到当前目录,可使用如下命令:
+```bash
+cp -r FastDeploy/build/fastdeploy-tmivx/ FastDeploy/examples/vision/classification/paddleclas/a311d/cpp/
+```
+
+3. 在当前路径下载部署所需的模型和示例图片:
+```bash
+mkdir models && mkdir images
+wget https://bj.bcebos.com/paddlehub/fastdeploy/ResNet50_vd_infer.tgz
+tar -xvf ResNet50_vd_infer.tgz
+cp -r ResNet50_vd_infer models
+wget https://gitee.com/paddlepaddle/PaddleClas/raw/release/2.4/deploy/images/ImageNet/ILSVRC2012_val_00000010.jpeg
+cp -r ILSVRC2012_val_00000010.jpeg images
+```
+
+4. 编译部署示例,可使入如下命令:
+```bash
+mkdir build && cd build
+cmake -DCMAKE_TOOLCHAIN_FILE=${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:
+
+
+
+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)
diff --git a/examples/vision/classification/paddleclas/a311d/cpp/infer.cc b/examples/vision/classification/paddleclas/a311d/cpp/infer.cc
new file mode 100755
index 000000000..140311eec
--- /dev/null
+++ b/examples/vision/classification/paddleclas/a311d/cpp/infer.cc
@@ -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
+#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;
+}
diff --git a/examples/vision/classification/paddleclas/a311d/cpp/run_with_adb.sh b/examples/vision/classification/paddleclas/a311d/cpp/run_with_adb.sh
new file mode 100755
index 000000000..aacaed4c5
--- /dev/null
+++ b/examples/vision/classification/paddleclas/a311d/cpp/run_with_adb.sh
@@ -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/$IMAGE_NAME"
diff --git a/examples/vision/classification/paddleclas/rv1126/cpp/README.md b/examples/vision/classification/paddleclas/rv1126/cpp/README.md
index feaba462f..b41fecace 100755
--- a/examples/vision/classification/paddleclas/rv1126/cpp/README.md
+++ b/examples/vision/classification/paddleclas/rv1126/cpp/README.md
@@ -32,7 +32,7 @@ cp -r ILSVRC2012_val_00000010.jpeg images
4. 编译部署示例,可使入如下命令:
```bash
mkdir build && cd build
-cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/timvx.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx ..
+cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=armhf ..
make -j8
make install
# 成功编译之后,会生成 install 文件夹,里面有一个运行 demo 和部署所需的库
diff --git a/examples/vision/classification/paddleclas/rv1126/cpp/infer.cc b/examples/vision/classification/paddleclas/rv1126/cpp/infer.cc
index c89510342..140311eec 100755
--- a/examples/vision/classification/paddleclas/rv1126/cpp/infer.cc
+++ b/examples/vision/classification/paddleclas/rv1126/cpp/infer.cc
@@ -48,7 +48,6 @@ int main(int argc, char* argv[]) {
if (argc < 3) {
std::cout << "Usage: infer_demo path/to/quant_model "
"path/to/image "
- "run_option, "
"e.g ./infer_demo ./ResNet50_vd_quant ./test.jpeg"
<< std::endl;
return -1;
diff --git a/examples/vision/detection/paddledetection/a311d/README.md b/examples/vision/detection/paddledetection/a311d/README.md
new file mode 100755
index 000000000..e5ba7376d
--- /dev/null
+++ b/examples/vision/detection/paddledetection/a311d/README.md
@@ -0,0 +1,11 @@
+# PP-YOLOE 量化模型在 A311D 上的部署
+目前 FastDeploy 已经支持基于 PaddleLite 部署 PP-YOLOE 量化模型到 A311D 上。
+
+模型的量化和量化模型的下载请参考:[模型量化](../quantize/README.md)
+
+
+## 详细部署文档
+
+在 A311D 上只支持 C++ 的部署。
+
+- [C++部署](cpp)
diff --git a/examples/vision/detection/paddledetection/a311d/cpp/CMakeLists.txt b/examples/vision/detection/paddledetection/a311d/cpp/CMakeLists.txt
new file mode 100755
index 000000000..7a145177e
--- /dev/null
+++ b/examples/vision/detection/paddledetection/a311d/cpp/CMakeLists.txt
@@ -0,0 +1,38 @@
+PROJECT(infer_demo C CXX)
+CMAKE_MINIMUM_REQUIRED (VERSION 3.10)
+
+# 指定下载解压后的fastdeploy库路径
+option(FASTDEPLOY_INSTALL_DIR "Path of downloaded fastdeploy sdk.")
+
+include(${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 ${ADB_TOOLS} DESTINATION ./)
diff --git a/examples/vision/detection/paddledetection/a311d/cpp/README.md b/examples/vision/detection/paddledetection/a311d/cpp/README.md
new file mode 100755
index 000000000..d0f4ff63a
--- /dev/null
+++ b/examples/vision/detection/paddledetection/a311d/cpp/README.md
@@ -0,0 +1,55 @@
+# PP-YOLOE 量化模型 C++ 部署示例
+
+本目录下提供的 `infer.cc`,可以帮助用户快速完成 PP-YOLOE 量化模型在 A311D 上的部署推理加速。
+
+## 部署准备
+### FastDeploy 交叉编译环境准备
+- 1. 软硬件环境满足要求,以及交叉编译环境的准备,请参考:[FastDeploy 交叉编译环境准备](../../../../../../docs/cn/build_and_install/a311d.md#交叉编译环境搭建)
+
+### 模型准备
+- 1. 用户可以直接使用由 FastDeploy 提供的量化模型进行部署。
+- 2. 用户可以先使用 PaddleDetection 自行导出 Float32 模型,注意导出模型模型时设置参数:use_shared_conv=False,更多细节请参考:[PP-YOLOE](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.4/configs/ppyoloe)
+- 3. 用户可以使用 FastDeploy 提供的[一键模型自动化压缩工具](../../../../../../tools/common_tools/auto_compression/),自行进行模型量化, 并使用产出的量化模型进行部署。(注意: 推理量化后的检测模型仍然需要FP32模型文件夹下的 infer_cfg.yml 文件,自行量化的模型文件夹内不包含此 yaml 文件,用户从 FP32 模型文件夹下复制此yaml文件到量化后的模型文件夹内即可。)
+- 更多量化相关相关信息可查阅[模型量化](../../quantize/README.md)
+
+## 在 A311D 上部署量化后的 PP-YOLOE 检测模型
+请按照以下步骤完成在 A311D 上部署 PP-YOLOE 量化模型:
+1. 交叉编译编译 FastDeploy 库,具体请参考:[交叉编译 FastDeploy](../../../../../../docs/cn/build_and_install/a311d.md#基于-paddlelite-的-fastdeploy-交叉编译库编译)
+
+2. 将编译后的库拷贝到当前目录,可使用如下命令:
+```bash
+cp -r FastDeploy/build/fastdeploy-tmivx/ FastDeploy/examples/vision/detection/yolov5/a311d/cpp
+```
+
+3. 在当前路径下载部署所需的模型和示例图片:
+```bash
+mkdir models && mkdir images
+wget https://bj.bcebos.com/fastdeploy/models/ppyoloe_noshare_qat.tar.gz
+tar -xvf ppyoloe_noshare_qat.tar.gz
+cp -r ppyoloe_noshare_qat models
+wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/000000014439.jpg
+cp -r 000000014439.jpg images
+```
+
+4. 编译部署示例,可使入如下命令:
+```bash
+mkdir build && cd build
+cmake -DCMAKE_TOOLCHAIN_FILE=${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:
+
+
+
+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)
diff --git a/examples/vision/detection/paddledetection/a311d/cpp/infer_ppyoloe.cc b/examples/vision/detection/paddledetection/a311d/cpp/infer_ppyoloe.cc
new file mode 100755
index 000000000..609a41d4b
--- /dev/null
+++ b/examples/vision/detection/paddledetection/a311d/cpp/infer_ppyoloe.cc
@@ -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;
+}
diff --git a/examples/vision/detection/paddledetection/a311d/cpp/run_with_adb.sh b/examples/vision/detection/paddledetection/a311d/cpp/run_with_adb.sh
new file mode 100755
index 000000000..dd7d7b47d
--- /dev/null
+++ b/examples/vision/detection/paddledetection/a311d/cpp/run_with_adb.sh
@@ -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/$IMAGE_NAME"
diff --git a/examples/vision/detection/paddledetection/rv1126/cpp/README.md b/examples/vision/detection/paddledetection/rv1126/cpp/README.md
index 5b366bd83..193a269fd 100755
--- a/examples/vision/detection/paddledetection/rv1126/cpp/README.md
+++ b/examples/vision/detection/paddledetection/rv1126/cpp/README.md
@@ -34,7 +34,7 @@ cp -r 000000014439.jpg images
4. 编译部署示例,可使入如下命令:
```bash
mkdir build && cd build
-cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/timvx.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx ..
+cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=armhf ..
make -j8
make install
# 成功编译之后,会生成 install 文件夹,里面有一个运行 demo 和部署所需的库
diff --git a/examples/vision/detection/paddledetection/rv1126/cpp/infer_ppyoloe.cc b/examples/vision/detection/paddledetection/rv1126/cpp/infer_ppyoloe.cc
index 77368584f..609a41d4b 100755
--- a/examples/vision/detection/paddledetection/rv1126/cpp/infer_ppyoloe.cc
+++ b/examples/vision/detection/paddledetection/rv1126/cpp/infer_ppyoloe.cc
@@ -53,7 +53,6 @@ int main(int argc, char* argv[]) {
if (argc < 3) {
std::cout << "Usage: infer_demo path/to/quant_model "
"path/to/image "
- "run_option, "
"e.g ./infer_demo ./PPYOLOE_L_quant ./test.jpeg"
<< std::endl;
return -1;
diff --git a/examples/vision/detection/yolov5/a311d/README.md b/examples/vision/detection/yolov5/a311d/README.md
new file mode 100755
index 000000000..d0bcd10a9
--- /dev/null
+++ b/examples/vision/detection/yolov5/a311d/README.md
@@ -0,0 +1,11 @@
+# YOLOv5 量化模型在 A311D 上的部署
+目前 FastDeploy 已经支持基于 PaddleLite 部署 YOLOv5 量化模型到 A311D 上。
+
+模型的量化和量化模型的下载请参考:[模型量化](../quantize/README.md)
+
+
+## 详细部署文档
+
+在 A311D 上只支持 C++ 的部署。
+
+- [C++部署](cpp)
diff --git a/examples/vision/detection/yolov5/a311d/cpp/CMakeLists.txt b/examples/vision/detection/yolov5/a311d/cpp/CMakeLists.txt
new file mode 100755
index 000000000..3c9eee38a
--- /dev/null
+++ b/examples/vision/detection/yolov5/a311d/cpp/CMakeLists.txt
@@ -0,0 +1,37 @@
+PROJECT(infer_demo C CXX)
+CMAKE_MINIMUM_REQUIRED (VERSION 3.10)
+
+# 指定下载解压后的fastdeploy库路径
+option(FASTDEPLOY_INSTALL_DIR "Path of downloaded fastdeploy sdk.")
+
+include(${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 ${ADB_TOOLS} DESTINATION ./)
diff --git a/examples/vision/detection/yolov5/a311d/cpp/README.md b/examples/vision/detection/yolov5/a311d/cpp/README.md
new file mode 100755
index 000000000..d47027bb0
--- /dev/null
+++ b/examples/vision/detection/yolov5/a311d/cpp/README.md
@@ -0,0 +1,54 @@
+# YOLOv5 量化模型 C++ 部署示例
+
+本目录下提供的 `infer.cc`,可以帮助用户快速完成 YOLOv5 量化模型在 A311D 上的部署推理加速。
+
+## 部署准备
+### FastDeploy 交叉编译环境准备
+- 1. 软硬件环境满足要求,以及交叉编译环境的准备,请参考:[FastDeploy 交叉编译环境准备](../../../../../../docs/cn/build_and_install/a311d.md#交叉编译环境搭建)
+
+### 量化模型准备
+- 1. 用户可以直接使用由 FastDeploy 提供的量化模型进行部署。
+- 2. 用户可以使用 FastDeploy 提供的[一键模型自动化压缩工具](../../../../../../tools/common_tools/auto_compression/),自行进行模型量化, 并使用产出的量化模型进行部署。
+- 更多量化相关相关信息可查阅[模型量化](../../quantize/README.md)
+
+## 在 A311D 上部署量化后的 YOLOv5 检测模型
+请按照以下步骤完成在 A311D 上部署 YOLOv5 量化模型:
+1. 交叉编译编译 FastDeploy 库,具体请参考:[交叉编译 FastDeploy](../../../../../../docs/cn/build_and_install/a311d.md#基于-paddlelite-的-fastdeploy-交叉编译库编译)
+
+2. 将编译后的库拷贝到当前目录,可使用如下命令:
+```bash
+cp -r FastDeploy/build/fastdeploy-tmivx/ FastDeploy/examples/vision/detection/yolov5/a311d/cpp
+```
+
+3. 在当前路径下载部署所需的模型和示例图片:
+```bash
+mkdir models && mkdir images
+wget https://bj.bcebos.com/fastdeploy/models/yolov5s_ptq_model.tar.gz
+tar -xvf yolov5s_ptq_model.tar.gz
+cp -r yolov5s_ptq_model models
+wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/000000014439.jpg
+cp -r 000000014439.jpg images
+```
+
+4. 编译部署示例,可使入如下命令:
+```bash
+mkdir build && cd build
+cmake -DCMAKE_TOOLCHAIN_FILE=${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:
+
+
+
+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)
diff --git a/examples/vision/detection/yolov5/a311d/cpp/infer.cc b/examples/vision/detection/yolov5/a311d/cpp/infer.cc
new file mode 100755
index 000000000..f1cf9e8dc
--- /dev/null
+++ b/examples/vision/detection/yolov5/a311d/cpp/infer.cc
@@ -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;
+}
diff --git a/examples/vision/detection/yolov5/a311d/cpp/run_with_adb.sh b/examples/vision/detection/yolov5/a311d/cpp/run_with_adb.sh
new file mode 100755
index 000000000..aacaed4c5
--- /dev/null
+++ b/examples/vision/detection/yolov5/a311d/cpp/run_with_adb.sh
@@ -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/$IMAGE_NAME"
diff --git a/examples/vision/detection/yolov5/rv1126/cpp/README.md b/examples/vision/detection/yolov5/rv1126/cpp/README.md
index 9711577f2..b974a8ebe 100755
--- a/examples/vision/detection/yolov5/rv1126/cpp/README.md
+++ b/examples/vision/detection/yolov5/rv1126/cpp/README.md
@@ -33,7 +33,7 @@ cp -r 000000014439.jpg images
4. 编译部署示例,可使入如下命令:
```bash
mkdir build && cd build
-cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/timvx.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx ..
+cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/toolchain.cmake -DFASTDEPLOY_INSTALL_DIR=${PWD}/../fastdeploy-tmivx -DTARGET_ABI=armhf ..
make -j8
make install
# 成功编译之后,会生成 install 文件夹,里面有一个运行 demo 和部署所需的库
diff --git a/examples/vision/segmentation/paddleseg/a311d/README.md b/examples/vision/segmentation/paddleseg/a311d/README.md
new file mode 100755
index 000000000..f65172cdd
--- /dev/null
+++ b/examples/vision/segmentation/paddleseg/a311d/README.md
@@ -0,0 +1,11 @@
+# PP-LiteSeg 量化模型在 A311D 上的部署
+目前 FastDeploy 已经支持基于 PaddleLite 部署 PP-LiteSeg 量化模型到 A311D 上。
+
+模型的量化和量化模型的下载请参考:[模型量化](../quantize/README.md)
+
+
+## 详细部署文档
+
+在 A311D 上只支持 C++ 的部署。
+
+- [C++部署](cpp)
diff --git a/examples/vision/segmentation/paddleseg/a311d/cpp/CMakeLists.txt b/examples/vision/segmentation/paddleseg/a311d/cpp/CMakeLists.txt
new file mode 100755
index 000000000..baaf8331f
--- /dev/null
+++ b/examples/vision/segmentation/paddleseg/a311d/cpp/CMakeLists.txt
@@ -0,0 +1,38 @@
+PROJECT(infer_demo C CXX)
+CMAKE_MINIMUM_REQUIRED (VERSION 3.10)
+
+# 指定下载解压后的fastdeploy库路径
+option(FASTDEPLOY_INSTALL_DIR "Path of downloaded fastdeploy sdk.")
+
+include(${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 ${ADB_TOOLS} DESTINATION ./)
diff --git a/examples/vision/segmentation/paddleseg/a311d/cpp/README.md b/examples/vision/segmentation/paddleseg/a311d/cpp/README.md
new file mode 100755
index 000000000..872784188
--- /dev/null
+++ b/examples/vision/segmentation/paddleseg/a311d/cpp/README.md
@@ -0,0 +1,54 @@
+# PP-LiteSeg 量化模型 C++ 部署示例
+
+本目录下提供的 `infer.cc`,可以帮助用户快速完成 PP-LiteSeg 量化模型在 A311D 上的部署推理加速。
+
+## 部署准备
+### FastDeploy 交叉编译环境准备
+- 1. 软硬件环境满足要求,以及交叉编译环境的准备,请参考:[FastDeploy 交叉编译环境准备](../../../../../../docs/cn/build_and_install/a311d.md#交叉编译环境搭建)
+
+### 模型准备
+- 1. 用户可以直接使用由 FastDeploy 提供的量化模型进行部署。
+- 2. 用户可以使用 FastDeploy 提供的一键模型自动化压缩工具,自行进行模型量化, 并使用产出的量化模型进行部署.(注意: 推理量化后的分类模型仍然需要FP32模型文件夹下的 deploy.yaml 文件, 自行量化的模型文件夹内不包含此 yaml 文件, 用户从FP32模型文件夹下复制此yaml文件到量化后的模型文件夹内即可.)
+- 更多量化相关相关信息可查阅[模型量化](../../quantize/README.md)
+
+## 在 A311D 上部署量化后的 PP-LiteSeg 分割模型
+请按照以下步骤完成在 A311D 上部署 PP-LiteSeg 量化模型:
+1. 交叉编译编译 FastDeploy 库,具体请参考:[交叉编译 FastDeploy](../../../../../../docs/cn/build_and_install/a311d.md#基于-paddlelite-的-fastdeploy-交叉编译库编译)
+
+2. 将编译后的库拷贝到当前目录,可使用如下命令:
+```bash
+cp -r FastDeploy/build/fastdeploy-tmivx/ FastDeploy/examples/vision/segmentation/paddleseg/a311d/cpp
+```
+
+3. 在当前路径下载部署所需的模型和示例图片:
+```bash
+mkdir models && mkdir images
+wget https://bj.bcebos.com/fastdeploy/models/rk1/ppliteseg.tar.gz
+tar -xvf ppliteseg.tar.gz
+cp -r ppliteseg models
+wget https://paddleseg.bj.bcebos.com/dygraph/demo/cityscapes_demo.png
+cp -r cityscapes_demo.png images
+```
+
+4. 编译部署示例,可使入如下命令:
+```bash
+mkdir build && cd build
+cmake -DCMAKE_TOOLCHAIN_FILE=${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:
+
+
+
+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)
diff --git a/examples/vision/segmentation/paddleseg/a311d/cpp/infer.cc b/examples/vision/segmentation/paddleseg/a311d/cpp/infer.cc
new file mode 100755
index 000000000..b6138e8fb
--- /dev/null
+++ b/examples/vision/segmentation/paddleseg/a311d/cpp/infer.cc
@@ -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;
+}
diff --git a/examples/vision/segmentation/paddleseg/a311d/cpp/run_with_adb.sh b/examples/vision/segmentation/paddleseg/a311d/cpp/run_with_adb.sh
new file mode 100755
index 000000000..aacaed4c5
--- /dev/null
+++ b/examples/vision/segmentation/paddleseg/a311d/cpp/run_with_adb.sh
@@ -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/$IMAGE_NAME"
diff --git a/examples/vision/segmentation/paddleseg/rv1126/cpp/README.md b/examples/vision/segmentation/paddleseg/rv1126/cpp/README.md
index 6295fa728..bf7cafc3b 100755
--- a/examples/vision/segmentation/paddleseg/rv1126/cpp/README.md
+++ b/examples/vision/segmentation/paddleseg/rv1126/cpp/README.md
@@ -33,7 +33,7 @@ cp -r cityscapes_demo.png images
4. 编译部署示例,可使入如下命令:
```bash
mkdir build && cd build
-cmake -DCMAKE_TOOLCHAIN_FILE=${PWD}/../fastdeploy-tmivx/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 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
diff --git a/examples/vision/segmentation/paddleseg/rv1126/cpp/infer.cc b/examples/vision/segmentation/paddleseg/rv1126/cpp/infer.cc
index 8c9c7456c..f084e6719 100755
--- a/examples/vision/segmentation/paddleseg/rv1126/cpp/infer.cc
+++ b/examples/vision/segmentation/paddleseg/rv1126/cpp/infer.cc
@@ -53,7 +53,6 @@ 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 ./ResNet50_vd_quant ./test.jpeg"
<< std::endl;
return -1;
diff --git a/fastdeploy/vision/detection/contrib/rknpu2/utils.h b/fastdeploy/vision/detection/contrib/rknpu2/utils.h
index 23efa25c8..1d28b5f0e 100644
--- a/fastdeploy/vision/detection/contrib/rknpu2/utils.h
+++ b/fastdeploy/vision/detection/contrib/rknpu2/utils.h
@@ -12,6 +12,7 @@
// See the License for the specific language governing permissions and
// limitations under the License.
#pragma once
+#include
#include
#include
diff --git a/tools/timvx/install.sh b/tools/timvx/install.sh
new file mode 100644
index 000000000..b499b09a1
--- /dev/null
+++ b/tools/timvx/install.sh
@@ -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