A jointly enabled solution by DeepScenario and Intel® SceneScape creates 4D digital twins of any parking lot using monocular cameras.
DeepScenario, Intel Ignite Europe batch 4 alumni, uses artificial intelligence to overcome the limitations of current image-based 2D solutions that are widely deployed in public spaces for security and smart city usages. The limitations include a lack of depth information or objects in the foreground occluding the view of objects behind them. DeepScenario delivers 3D computer vision for off-the-shelf 2D cameras, preventing the cost-prohibitive exercise of replacing all the cameras in the system with 3D sensors like LIDARs.
The foundational software is capable of detecting and tracking all moving objects in the scene, such as people, vehicles and equipment, then outputting them in 3D space with accuracy in the centimeter range. As an example of using this technology in conjunction with Intel® SceneScape, DeepScenario has developed an advanced smart parking solution.
Based on video inputs, DeepScenario’s model recognizes vehicles in 3D space, categorizes them by type (car, van, truck, etc.) and extracts their 3D bounding boxes and trajectories. This capability is used to determine the position, orientation, and dimension of all vehicles on the parking lot with highest accuracy.
Through their collaboration, DeepScenario and Intel are pioneering new capabilities that combine DeepScenario’s AI-powered computer vision software with the Intel® SceneScape spatial awareness capabilities delivered with the Intel® Tiber™ Edge Platform. To demonstrate this combination, a smart parking usage model — with deep extensibility to other use cases — provides capabilities such as parking-violation detection and parking lot occupancy monitoring. Linking the physical and digital world with DeepScenario.
