Detectron2 Sample Deployment

Why this

Deploying a detectron based RNN model is difficult. In spite of the offical guide of deployment, it's hard to compile the environment for caffe2 or stuff like that. I know little about C++ and Linux, and most of the ML frameworks are not supporting Windows. Anyway, My app is simple and I'm not worrying about the performance. I just want it works, and I'm saving my time.

What's this

This is an interface for other programs to get the predict data of the detectron2 based RNN detecter.

usage

The model file should be put in engine/output/model_final.pth, which is not included here.

Install Docker.

 $ git clone https://github.com/CoccaGuo/Detectron2-Simple-Deployment.git
 $ cd Detectron2-Simple-Deployment

$ mkdir engine/output/ app/static/upload/ app/static/result/

$ docker build -t cocca/co_detect

$ docker run -d -p 5000:5000 cocca/co_detect

Open your browser and goto localhost:5000, or labview open client.

What's more

This project may only works with this model. Act with mildly caution.