This repo contains the code to my master degree thesis entitled VIDEO ANALYTICS FOR CONNECTION DISCOVER AND USER BEHAVIOR MONITORING,
The main objective of the thesis was to discover connections between customers at a retail store such as coffee shop or a small cafe based on the behaviors they exhibit at the store. All these using the surveillance camera at the store.
I use state-of-the-art machine vision and deep learning algorithms to
- recognize people,
- track them,
- classify their actions as whether they are sitting or standing
- calculate distance between them
- and if the distance is within a certain threshold, determine their gaze direction
- finally categorize them into a group if they satisfy the given conditions.
The customers in yellow boxes are in a group.