{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/0ef4d2b3e2004c9c8a03e88cbd1b3577\" frameborder=\"0\" width=\"1660\" height=\"1245\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1245,"width":1660,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1245,"thumbnail_width":1660,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/0ef4d2b3e2004c9c8a03e88cbd1b3577-2edff22d2dd0b2c8.gif","duration":932.612,"title":"AI-Driven Visitor Counting System Using Video Streams","description":"Hi, I'm Vithahaselvi Haribalajhee, and in this video, I present my AI-driven system designed to count unique visitors in video streams, whether from live cameras or pre-recorded footage. I utilized YOLO for face detection, tracking, and recognition, and implemented a multi-stage pipeline consisting of detection, recognition, tracking, logging, and counting. I faced challenges with false positives and multi-embedding, which I addressed through various techniques, including a confirmation buffer and threshold calibration. I also demonstrate the system's functionality with a live demo, showing that it successfully identified eight unique visitors. I encourage you to review the GitHub repository for more details on the implementation."}