paddlepaddle检测模PC端服务端部署CPU或GPU傻瓜步骤

目标检测训练好的模型在服务端生成可调用的接口

链接奉上:https://paddlex.readthedocs.io/zh_CN/develop/deploy/hub_serving.html

准备工作：

pip install paddlehub==1.8.0

pip install paddlepaddle==1.8.0

pip install paddlex

按步骤来(已训练好且转换的模型是fast101_inference：

 hub convert --model_dir /home/lsj/Videos/fastrcnn__101/fast101_inference --module_name fast101_inference

The converted module is stored in fast101_inference_1597226125.3915954（在你的cmd端同路径下哦

模型安装生成的模型里面的tar.gz：

hub install /home/lsj/Videos/fastrcnn__101/fast101_inference_1597226125.3915954/fast101_inference.tar.gz

The converted module is stored in `fast101_inference_1597226125.391595`.