Hi! I am a fourth-year B.Tech Computer Science student with a specialization in Computer Vision at the International Institute of Information Technology, Hyderabad, under the guidance of Dr. Ravi Kiran.

At CVIT I am part of the mobility group led by Dr. Ravi Kiran, where I work on building traffic violation detection systems using dashcam footage from moving vehicles. Unlike traditional autonomous driving approaches, our focus is on unstructured, dense environments.

We have developed a novel methodology for detecting triple riding and helmet violations on high-density Indian roads using only monocular dashcam video. This is the first approach addressing such violations from a moving vehicle, overcoming challenges like high occlusion, rider-to-motorcycle ambiguity, and difficulty in tracking vehicles for license plate detection.

Our work introduces:

• A novel cross-segmentation algorithm
• A novel multiple-object tracking method
• A large scale video dataset annotated with bounding boxes, track IDs, helmet, and license plate information

The work was accepted to WACV 2025. I am also working on detecting Wrong Side Driving on Indian roads.

In parallel, I am conducting an Independent Study at our college’s Robotics Research Center (RRC) under K. Madhava Krishna. This project aims to detect free-space drivable areas using just an RGB image, tackling the challenge with minimal annotated data, i.e., ego motion and the RGB front camera data.

Additionally, I am collaborating with Dr. Charu Sharma and Dr. Avinash Sharma on Gaussian splatting for 3D scene reconstruction and novel view synthesis. We are exploring interpretable compression techniques by identifying local repetitions and symmetries in the scene for efficient storage.