jesseallardice/action_recording

Action_Recording

This project enables the user to easily collect a automatically annotated dataset of human exercise with video. The collected datasets can then be used to train a deep learning model for exercise recognition and repetition rate estimation with the project Action-Recognition.

scheduleCap.py

scheduleCap.py collects a labelled dataset of exercise movements in the form of ordered webcam images.

Before running scheduleCap.py

• check that the testing is off (TESTING = False)
• that the user id (USER_ID) is correct

Workout schedule

information is stored and retreved from the file: Action_Recording_Schedule.json

This includes information such as action_types, a list of possible actions state_types, a list of possible states in the schedule. schedule, list of entries in the form {state: time period}