VECID software package
EyeVision software + Deep Learning modul
Patterns are taught into a neural network and then recognized again, even if they appear in front of the camera in different positions or partially destroyed. This guarantees detection even under less than ideal conditions. The neural networks are fatigue-free and always extremely precise. If the human brain throttles the recognition performance significantly after 15 minutes, the EyeVision software works just as precisely as before.
The recognition of the license plates and the reading of the characters can be optimally solved with the Deep Learning component of the EyeVision 4 software. Of course, all cars have number plates, but they are not always fixed in the same place or have different character sets. In such an application, the human recognizes errors that occur immediately, whereas classic image processing does not. However, deep learning offers the solution to this problem.
Vecid enables the number plates to be recognized and the individual components to be read. In addition, the exact determination of the license plate makes it possible to make certain parts illegible.
The VECID system runs on all hardware platforms. From a simple embedded system with an ARM processor to a high-end PC with the latest CPU and GPU. The difference lies in the evaluation speed with which the vehicles are recognized. A cloud solution is also available for distributed systems, which runs on an AWS basis, but can also be made available for other servers.
The integration of all standard cameras allows scaling from inexpensive ARM MIPI camera systems with a few readings per second to high-end PCs with multi-camera support and several hundred readings per second. A complete integration of the software into the communication systems allows all possible business models from Pay Per Read to a dedicated computer system to be implemented.
To determine the optimal hardware/software combination for your application, please contact us by phone or e-mail.