LIVE ACCIDENT DETECTION AND ALERT SYSTEM
- Version
- Download 8
- File Size 494.58 KB
- File Count 1
- Create Date 19 May 2025
- Last Updated 19 May 2025
LIVE ACCIDENT DETECTION AND ALERT SYSTEM
Authors:
P. Prashamsa1, Mula Ruthik Reddy2, Kothapally Sai Ram3, Nandipati Kalyan4
1-4 Department of CSE & TKR College of Engineering & Technology
2-5 B.Tech Students
ABSTRACT: Accident detection in real-time is a crucial aspect of intelligent traffic monitoring systems aimed at improving road safety. This project introduces a deep learning-based live accident detection and alert system using a Convolutional Neural Network (CNN) integrated with bounding box overlap logic. The system utilizes TensorFlow’s object detection API to identify vehicles from live webcam footage and determines collisions based on spatial intersection of detected objects. It operates without physical sensors or motion hardware, making it lightweight and cost- efficient. Location of the detected accident is retrieved using IP- based geolocation, and an alert is generated through a Flask web interface. The model demonstrates reliable accuracy and fast response time, making it well-suited for smart city surveillance, urban monitoring, and real-time emergency alert applications.
Keywords — Live Accident Detection, Convolutional Neural Network (CNN), Object Detection, Emergency Alert System, Smart City Surveillance