Early Achievement: Computer Vision for Intelligent Traffic Monitoring
Developed a computer vision pipeline for the Omdena Bangladesh Chapter challenge using YOLOv4 for multi-class vehicle detection and DeepSORT for object tracking. The system detects 9 vehicle classes and visualizes relative vehicle velocity in real time (green for positive, red for negative motion). Despite limited time and an imbalanced dataset, the project demonstrated a strong proof of concept with scope for improved accuracy through future data collection and model refinement.




