This was a course project for the course Advanced Vision, part of my MSc. Artificial Intelligence program at University of Edinburgh.
The objective of this project was to build a system for detecting motion in an indoor environment, and for tracking detected objects
based on their color distribution. The input video feed was captured using a color-camera, mounted on a stable platform.
A background image of the scene (without moving objects) was used as a reference frame. A simple pixel subtraction between the
background image and subsequent image frames was done for detecting motion. Detected moving objects were classified and assigned
ids based on the color information. In subsequent image frames, detected moving objects that matched closest (measured as
Bhattacharyya distance) to previously (in the previous frame)
classified objects were assigned the same ids respectively, and used as reference object for the next frame.