VEHICLE DETECTION AND SPEED DETECTION USING YOLOV8

Abstract

This project presents a comprehensive framework for efficient object detection and speed estimation within video streams, employing the YOLOv8n model and associated libraries. The YOLOv8n model is loaded and systematically applied to identify objects in each frame, with annotated frames compiled into a new video sequence. The subsequent speed detection process relies on consecutive frame analysis to estimate vehicle velocities. By tracking vehicles through bounding boxes and computing their displacement using the Eu- clidean distance formula, the system accurately measures spatial movement. Integration of a calibration factor converts pixel-based measurements to real-world units, and a con- version factor transforms speed to kilometers per hour. The resulting speed estimations are superimposed onto bounding boxes in the video frames, providing a visually enhanced representation of vehicle speeds and movements. This adaptable and robust methodology offers valuable insights for applications requiring object detection and speed detection.

Keywords:  YOLOv8, Object Detection, Speed Detection, Ultralytics