Parking Occupancy Prediction using Computer Vision with Location Awareness

Parking occupancy in large outdoor parking as well as at street parking slots is a big problem, and solutions for it were proposed in the past with limited success. Recent advances in Computer Vision (CV) algorithms made this technology more powerful by making models smaller and more precise. On the other hand development of hardware components made Edge devices more powerful with AI accelerators as built-in options. Based on this new development, we propose an architecture where the latest Smart Edge devices are used to calculate parking occupancy on the edge. The proposed solution uses newly available data streams to create a 3D model of the parking environment that will identify available on-street parking spaces with high precision.