Publication Details

Abstract

This project introduces a web-based application for real-time car and pedestrian detection in video streams using the YOLOv5s object detection model. Developed with the Streamlit framework, the system offers a user-friendly interface that supports both live webcam feeds and uploaded video files. By leveraging the lightweight and efficient YOLOv5s model from the Ultralytics repository, the application delivers high-speed detection with practical accuracy, suitable for real-time use. Key features include real-time visualization, interactive controls, and downloadable output, making it well-suited for use cases such as traffic monitoring, pedestrian safety, and urban planning.