Sahil Beniwal

At TCI Group, I developed YOLO-based truck rear detection and camera pipelines for dock occupancy monitoring. I integrated gate and ANPR events, deployed services on AWS, and improved observed ANPR response time from 6–8 seconds to 2–3 seconds through GPU inference and plate validation.

I also worked on bale and material detection, calibrated volume estimation, and expanded a BLE worker movement monitoring solution to 300 workers in a 500,000 sq ft warehouse. I built authenticated edge gateway access for private camera feeds, including approximately 30 feeds running at 1–2 FPS per stream on Raspberry Pi.

My research prototype explored multi-camera person tracking and cross-camera identity association using YOLO, McByte, and OSNet appearance embeddings; reliable warehouse-wide identity persistence remained under research and validation. At OneStepGreener, I worked on waste-image annotation, evaluated GroundingDINO and SAM2, and developed face-attendance anti-spoofing and hand-movement counting prototypes.