Vehicle Flow Analysis with Custom YOLO NAS in Antananarivo CCTV
This project involves training a custom YOLO NAS model to detect, classify, and count vehicles in CCTV footage from Antananarivo. The aim is to analyze traffic flow for urban planning and congestion management. Data and Labelling Data Type: CCTV video data capturing urban traffic. Labelling Types: Detection of vehicle locations. Classification of vehicle types (e.g., car, bus, motorcycle). Counting for flow analysis. Tools: Roboflow for automated and bulk annotation. LabelImg for manual refinement of bounding boxes and labels. Methodology Annotated video frames to identify vehicles and their categories. Trained the YOLO NAS model on the labelled dataset. Processed outputs to count vehicles per segment for flow analysis. Objectives Provide accurate vehicle counts for traffic flow insights. Support integration into dashboards or optimization tools