License Plate Recognition

The limitation of existing ALPR systems, is that they fail to adapt to number plates varying due to region or location. Custom numbers plates are another example where existing systems fail to deliver results.

The solution consists of a computer vision based approach that would detect, read and identify number plates of moving vehicles using CCTV cameras and then store them into a centralized database. This system can be deployed in restricted and high security areas or gated communities to monitor the movement of vehicles entering, exiting or inside the premises.

ByteCorp developed an inhouse ALPR system using custom deep learning models trained on a large diverse dataset, coupled with a unique approach for a single solution that caters to all types of Number Plates including custom number plates.

The solution would detect, read and identify number plates of varying sizes, locations and fonts and save them into a system. The system can be deployed on the edge for ease of scalability and cost reduction.