[Model] add vsr serials models (PaddlePaddle#518)

* [Model] add vsr serials models Signed-off-by: ChaoII <[email protected]> * [Model] add vsr serials models Signed-off-by: ChaoII <[email protected]> * fix build problem Signed-off-by: ChaoII <[email protected]> * fix code style Signed-off-by: ChaoII <[email protected]> * modify according to review suggestions Signed-off-by: ChaoII <[email protected]> * modify vsr trt example Signed-off-by: ChaoII <[email protected]> * update sr directory * fix BindPPSR * add doxygen comment * add sr unit test * update model file url Signed-off-by: ChaoII <[email protected]> Co-authored-by: Jason <[email protected]>
Zzzz-yh · Nov 21, 2022 · c7ec14d · c7ec14d
1 parent 1ac54c9
commit c7ec14d
Show file tree

Hide file tree

Showing 40 changed files with 2,526 additions and 8 deletions.
diff --git a/.gitignore b/.gitignore
@@ -39,4 +39,6 @@ coverage
 yalc.*
 .yalc
 examples/vision/collect_quantize_cc.sh
-examples/vision/tests_quantize
+examples/vision/tests_quantize
+fastdeploy/LICENSE
+fastdeploy/ThirdPartyNotices.txt
diff --git a/examples/vision/sr/README.md b/examples/vision/sr/README.md
@@ -0,0 +1,9 @@
+# sr 模型部署
+
+FastDeploy目前支持如下超分模型部署
+
+| 模型                                       | 说明                    | 模型格式   | 版本                                                                                |
+|:-----------------------------------------|:----------------------|:-------|:----------------------------------------------------------------------------------|
+| [PaddleGAN/BasicVSR](./basicvsr)         | BasicVSR 系列模型         | paddle | [develop](https://github.com/PaddlePaddle/PaddleGAN/blob/develop/docs/zh_CN/tutorials/video_super_resolution.md)                        |
+| [PaddleGAN/EDVR](./edvr)                 | EDVR 系列模型             | paddle | [develop](https://github.com/PaddlePaddle/PaddleGAN/blob/develop/docs/zh_CN/tutorials/video_super_resolution.md) |
+| [PaddleGAN/PP-MSVSR](./ppmsvsr)          | PP-MSVSR 系列模型         | paddle | [develop](https://github.com/PaddlePaddle/PaddleGAN/blob/develop/docs/zh_CN/tutorials/video_super_resolution.md) |
diff --git a/examples/vision/sr/basicvsr/README.md b/examples/vision/sr/basicvsr/README.md
@@ -0,0 +1,28 @@
+# BasicVSR模型部署
+
+## 模型版本说明
+
+- [PaddleGAN develop](https://github.com/PaddlePaddle/PaddleGAN)
+
+## 支持模型列表
+
+目前FastDeploy支持如下模型的部署
+
+- [BasicVSR](https://github.com/PaddlePaddle/PaddleGAN/blob/develop/docs/zh_CN/tutorials/video_super_resolution.md)。
+
+
+## 导出部署模型
+
+在部署前，需要先将训练好的BasicVSR导出成部署模型，导出BasicVSR导出模型步骤，参考文档[导出模型](https://github.com/PaddlePaddle/PaddleGAN/blob/develop/docs/zh_CN/tutorials/video_super_resolution.md)。
+
+
+| 模型                                                                          | 参数大小   | 精度    | 备注 |
+|:----------------------------------------------------------------------------|:-------|:----- | :------ |
+| [BasicVSR](https://bj.bcebos.com/paddlehub/fastdeploy/BasicVSR_reds_x4.tgz) | 30.1MB | - |
+
+**注意**：非常不建议在没有独立显卡的设备上运行该模型
+
+## 详细部署文档
+
+- [Python部署](python)
+- [C++部署](cpp)
diff --git a/examples/vision/sr/basicvsr/cpp/CMakeLists.txt b/examples/vision/sr/basicvsr/cpp/CMakeLists.txt
@@ -0,0 +1,14 @@
+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})
+
+add_executable(infer_demo ${PROJECT_SOURCE_DIR}/infer.cc)
+# 添加FastDeploy库依赖
+target_link_libraries(infer_demo ${FASTDEPLOY_LIBS})
diff --git a/examples/vision/sr/basicvsr/cpp/README.md b/examples/vision/sr/basicvsr/cpp/README.md
@@ -0,0 +1,74 @@
+# BasicVSR C++部署示例
+
+本目录下提供`infer.cc`快速完成BasicVSR在CPU/GPU，以及GPU上通过TensorRT加速部署的示例。
+
+在部署前，需确认以下两个步骤
+
+- 1. 软硬件环境满足要求，参考[FastDeploy环境要求](../../../../../docs/cn/build_and_install/download_prebuilt_libraries.md)
+- 2. 根据开发环境，下载预编译部署库和samples代码，参考[FastDeploy预编译库](../../../../../docs/cn/build_and_install/download_prebuilt_libraries.md)
+以Linux上BasicVSR推理为例，在本目录执行如下命令即可完成编译测试（如若只需在CPU上部署，可在[Fastdeploy C++预编译库](../../../../../docs/cn/build_and_install/download_prebuilt_libraries.md/CPP_prebuilt_libraries.md)下载CPU推理库）
+
+```bash
+#下载SDK，编译模型examples代码（SDK中包含了examples代码）
+# fastdeploy版本 >= 0.7.0
+wget https://bj.bcebos.com/fastdeploy/release/cpp/fastdeploy-linux-x64-gpu-0.7.0.tgz
+tar xvf fastdeploy-linux-x64-gpu-0.7.0.tgz
+cd fastdeploy-linux-x64-gpu-0.7.0/examples/vision/sr/basicvsr/cpp/
+mkdir build && cd build
+cmake .. -DFASTDEPLOY_INSTALL_DIR=${PWD}/../../../../../../../fastdeploy-linux-x64-gpu-0.7.0
+make -j
+
+# 下载BasicVSR模型文件和测试视频
+wget https://bj.bcebos.com/paddlehub/fastdeploy/BasicVSR_reds_x4.tar
+tar -xvf BasicVSR_reds_x4.tar
+wget https://bj.bcebos.com/paddlehub/fastdeploy/vsr_src.mp4
+
+
+# CPU推理
+./infer_demo BasicVSR_reds_x4 vsr_src.mp4 0 2
+# GPU推理
+./infer_demo BasicVSR_reds_x4 vsr_src.mp4 1 2
+# GPU上TensorRT推理
+./infer_demo BasicVSR_reds_x4 vsr_src.mp4 2 2
+```
+
+以上命令只适用于Linux或MacOS, Windows下SDK的使用方式请参考:  
+- [如何在Windows中使用FastDeploy C++ SDK](../../../../../docs/cn/faq/use_sdk_on_windows.md)
+
+## BasicVSR C++接口
+
+### BasicVSR类
+
+```c++
+fastdeploy::vision::sr::BasicVSR(
+        const string& model_file,
+        const string& params_file = "",
+        const RuntimeOption& runtime_option = RuntimeOption(),
+        const ModelFormat& model_format = ModelFormat::PADDLE)
+```
+
+BasicVSR模型加载和初始化，其中model_file为导出的Paddle模型格式。
+
+**参数**
+
+> * **model_file**(str): 模型文件路径
+> * **params_file**(str): 参数文件路径
+> * **runtime_option**(RuntimeOption): 后端推理配置，默认为None，即采用默认配置
+> * **model_format**(ModelFormat): 模型格式，默认为Paddle格式
+
+#### Predict函数
+
+> ```c++
+> BasicVSR::Predict(std::vector<cv::Mat>& imgs, std::vector<cv::Mat>& results)
+> ```
+>
+> 模型预测接口，输入图像直接输出检测结果。
+>
+> **参数**
+>
+> > * **imgs**: 输入视频帧序列，注意需为HWC，BGR格式
+> > * **results**: 视频超分结果，超分后的视频帧序列
+
+- [模型介绍](../../)
+- [Python部署](../python)
+- [如何切换模型推理后端引擎](../../../../../docs/cn/faq/how_to_change_backend.md)