Full Project Name: AI-integrated Surveillance System to Support Detecting Social Distancing & Face Mask Violations.
COVID-19 Violation Detection Assistance System (VDAS) is a graduation/capstone project done by a 4-student group (Professor Mafia) majored in Software Engineering at Can Tho FPT University.
VDAS aims to utilize Deep Learning to help authorities and organizations to easier manage, delineate pre-defined areas with the most COVID-19 violations (social distancing and face mask), and take preemptive actions.
VDAS provides the management capabilities for roles, cameras, areas, managers, staff, as well as the capabilities to detect face mask and social distancing violations, and generates relevant reports.
VDAS makes use of the YOLOv5 object detection model, with specific focus on dealing with low-light conditions using a proposed version of Local Binary Pattern - Sigmoid Local Pattern. The final product consists of 2 major parts: The web application, and the trained YOLOv5 model.
- Detect face mask/social distancing violations from image/video/camera stream
- Manage areas, cameras, roles, managers, and staff
- Export violation rate/specific violation case reports
- Sigmoid Local Pattern for Robust Car and Pedestrian Detection
- Robust Face Mask Detection Using Local Binary Pattern and Deep Learning
- mafia-web: Minh Thắng
- mafia-api: Minh Thắng, Quang Huy
- mafia-ai: Đức Lộc
- mafia-detector: Hoàng Phúc
NOTE: Some files' URLs are inaccessible due to being stored on terminated Edu Accounts.