Repository storing work toward detecting human activity states. Includes several tutorials I've found useful.
HAR focuses on detecting the probable activity that a human is performing from either a static image or a set of images in squence.
A HAR application is commonly divided into three stages:
- Detection of the human object
- detection of the human object's atamony or interest points *(used to determine the activity)*
- prediction of the activity based on the provided data/li>
Additional points of interest for research include the continuing squence of actions, next probable actions, actions as part of a group, detecting group behavior and many more.
To run this code, download the data folder and its contents and activate the notebook you wish to view. A good place to start is the main.ipynb