Dwarkadas J. Sanghvi College of Engineering (DJCSE)

• Developed a novel key-frame extraction algorithm for video surveillance system & optimized the run time by 97.5% • Employed a custom-trained YOLOv8 model with an accuracy of 95% to detect anomalies like fire and weapons • Authored a paper about video analysis and performed a thorough performance analysis for object detection models • Created a custom YOLO fire & weapon dataset specifically for detection from CCTV footage with low resolutions