I engineered a dual-YOLO accident detection system that identifies vehicle collisions from CCTV feeds and classifies their severity. I added temporal verification to reduce false positives and spatial analysis to find incident coordinates.
The project’s REST API dispatch pipeline routes alerts and navigation coordinates to the nearest available ambulance, with emergency notifications sent through Fast2SMS.
For a missing-child identification project, I developed a CNN application using TensorFlow and Keras to match facial features across datasets. It achieved 95% accuracy and reduced manual verification time by approximately 50%.
During my internships, I implemented computer vision and deep learning pipelines at InternPe, and built backend services using Java, SQL, and REST APIs at Jspiders. I also developed a student management application with Spring Boot and a home food service website.
