Javascript project that loads a fastreid model using tensorflow js with webgpu support. The model wrote in this repo is the ResNet50 custom architecture based on Bag of tricks: BoT(R50)
This project can be installed and tested through npm and webpack config:
$ npm install
$ npm startThe main entry points are in ./src/index.html and ./src/index.ts. The repo is ready to load the binary model and test images from an http server. You can run this other server cloning the model server repo.