Smart crowd monitoring systems
Project Description

This project uses a combination of image analysis tools ranging from rule-base, adaptive, statistical, and deep learning networks to segment, classify and estimate the movement of humans and their bodies as captured by surveillance videos. Crowd management is vital for public safety and the current challenges are the very poor quality of images / videos available.

Supervisor
SO Richard Hau Yue
Quota
3
Course type
UROP1100
UROP2100
UROP3100
Applicant's Roles

The selected candidate will work alongside a team of PhD and MPhil students on the above said project. He or she will be responsible for setting up an experimental test rig to capture the videos and develop codes based upon the established libraries.

Applicant's Learning Objectives

1. Ability to precisely control and manipulate optical geometry through lens manipulation;
2. Ability to develop codes to segment human features
3. Ability to integrate hardware and software together within the scope of real-time crowd monitoring

Complexity of the project
Challenging