Digit Recognition model

It is the base Convnet model used in the Digit recognition app for android

How it works

• The images from the server are stored in the to_predict folder. Only the first image is used for prediction so make sure that the folder contains only one image at all time which is to be processed.
• Image is processed by the model
• The results are stored in predict.json file. Top-3 predictions are returned as result in a JSON format file.

It can be implemented in any app in backend server.

1. Set the image path to to_predict folder
2. From inside your program call predict.py
3. The results are stored in pred.json
4. Pass the results back to your program from the json file.