mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2025-10-06 17:17:14 +08:00
cfb0a983ea
* Imporve OCR Readme * Improve OCR Readme * Improve OCR Readme * Improve OCR Readme * Improve OCR Readme * Add Initialize function to PP-OCR * Add Initialize function to PP-OCR * Add Initialize function to PP-OCR * Make all the model links come from PaddleOCR * Improve OCR readme * Improve OCR readme * Improve OCR readme * Improve OCR readme * Add Readme for vision results * Add Readme for vision results * Add Readme for vision results * Add Readme for vision results * Add Readme for vision results * Add Readme for vision results * Add Readme for vision results * Add Readme for vision results * Add Readme for vision results * Add Readme for vision results * Add check for label file in postprocess of Rec model * Add check for label file in postprocess of Rec model * Add check for label file in postprocess of Rec model * Add check for label file in postprocess of Rec model * Add check for label file in postprocess of Rec model * Add check for label file in postprocess of Rec model * Add comments to create API docs * Improve OCR comments * Rename OCR and add comments * Make sure previous python example works * Make sure previous python example works * Fix Rec model bug * Fix Rec model bug * Fix rec model bug * Add SetTrtMaxBatchSize function for TensorRT * Add SetTrtMaxBatchSize Pybind * Add set_trt_max_batch_size python function * Set TRT dynamic shape in PPOCR examples * Set TRT dynamic shape in PPOCR examples * Set TRT dynamic shape in PPOCR examples * Fix PPOCRv2 python example * Fix PPOCR dynamic input shape bug * Remove useless code * Fix PPOCR bug * Remove useless comments in PaddleSeg example Co-authored-by: Jason <jiangjiajun@baidu.com>
PaddleSeg 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PaddleSeg量化模型在CPU上的部署推理加速.
部署准备
FastDeploy环境准备
-
- 软硬件环境满足要求，参考FastDeploy环境要求
-
- FastDeploy Python whl包安装，参考FastDeploy Python安装
量化模型准备
-
- 用户可以直接使用由FastDeploy提供的量化模型进行部署.
-
- 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.)
以量化后的PP_LiteSeg_T_STDC1_cityscapes模型为例, 进行部署
在本目录执行如下命令即可完成编译,以及量化模型部署.支持此模型需保证FastDeploy版本0.7.0以上(x.x.x>=0.7.0)
mkdir build
cd build
# 下载FastDeploy预编译库，用户可在上文提到的`FastDeploy预编译库`中自行选择合适的版本使用
wget https://bj.bcebos.com/fastdeploy/release/cpp/fastdeploy-linux-x64-x.x.x.tgz
tar xvf fastdeploy-linux-x64-x.x.x.tgz
cmake .. -DFASTDEPLOY_INSTALL_DIR=${PWD}/fastdeploy-linux-x64-x.x.x
make -j
#下载FastDeloy提供的PP_LiteSeg_T_STDC1_cityscapes量化模型文件和测试图片
wget https://bj.bcebos.com/paddlehub/fastdeploy/PP_LiteSeg_T_STDC1_cityscapes_without_argmax_infer_PTQ.tar
tar -xvf PP_LiteSeg_T_STDC1_cityscapes_without_argmax_infer_PTQ.tar
wget https://paddleseg.bj.bcebos.com/dygraph/demo/cityscapes_demo.png
# 在CPU上使用Paddle-Inference推理量化模型
./infer_demo PP_LiteSeg_T_STDC1_cityscapes_without_argmax_infer_PTQ cityscapes_demo.png 1