Advanced Multi-camera vehicle identification, tracking and geo-localization system
The proposed project proposes design solutions for a Multi-camera vehicle identification, tracking and geo-localization system. The systems is constrained to use video streams from inexpensive surveillance cameras, and leverages image recognition software trained to detect and track vehicles in a controlled area. This will allow apps to pinpoint vehicles easily and reliably. Such technology works well in controlled environments but faces serious challenges when deployed in the real world. Anisotropic image backgrounds, distorted and noisy camera images, dirty vehicles, and difficult meteorological conditions make object recognition and object tracking more difficult. The artificial intelligence modules will be part of a robust distributed platform that offers strong back-end services and user-friendly interfaces, to support many sites and many cameras.