Skip to content

Latest commit

 

History

History
23 lines (20 loc) · 1.07 KB

README.md

File metadata and controls

23 lines (20 loc) · 1.07 KB
Raw

DIGIT_train

This is a repository including a sample code for creating a custom touch detection model for a DIGIT sensor.
See also: PyTouch

Requirements

It requires pytouch, digit-interface, torch, torchvision to be installed in your python.
Install with pip install if not.

Usage

1. Make a dataset

  1. Change the directory name Dxxxx to your DIGIT id in DIGIT_train/Dxxxx
  2. Change DIGIT_ID and the path to yours in make_data.py
  3. Run make_data.py to make your dataset in ./Dxxxx/touch and ./Dxxxx/notouch
    Each frame will be saved to the specified directory while running.
    Make sure to create various touching situations. To achieve this, I changed the touching situation for each set of 200 or 300 frames.

2. Train a model

Change DIGIT_ID in model.py and run it. .pth file will be saved under the ./weight directory.

3. Test the model

Change DIGIT_ID in realtime_touchdetect.py and run it.