christoz / ml-tensorflow-node

Development

# start in development mode using nodemon
yarn dev

# start in debug mode
yarn debug

# run linters (eslint + prettier)
yarn lint

# create production build
yarn build

# start production build
yarn start

Create human readable labels

NODE_LABEL=your_subject yarn createLabels

Upload data

The data are expected to be organised in folders matching the label which describes the class of the data subsequently for example it communicates as food type.

food/
-- foodA/
  -- IMG-1.jpg
  -- IMG-2.jpg
  -- IMG-3.jpg
-- foodB/
  -- IMG-1.jpg
  -- IMG-2.jpg
  -- IMG-3.jpg
  ...

zip the folder you've just created, open postman or somthing equivelant, add form data field "file" and select your zip file then request with the url:

POST: http://localhost:3000/api/classification/upload

Note the name of the zip file defines also the name of the model, so try to be picky

Classify data and save the model

GET: http://localhost:3000/api/classification/classify?subject=food

Download model

GET: http://localhost:3000/api/classification/download?file=food