Client
A corporate entity responsible for traffic safety.
Challenge
The client approached our development team with the task of building and implementing a set of neural networks to search and identify car number plates in a live video stream.
Solution
Our machine learning development team developed a set of neural networks that solved the search and recognition problem of identifying car number plates in a live video stream. The goal was to achieve consistent, high-quality results with more than 90% accuracy and work with a 4k video stream at 60 FPS or higher. In addition, the software was required to operate on an embedded computer, NVIDIA Jetson TX2. This required maximum acceleration and optimization of neural network calculations. Our development team built and optimized the software and was responsible for collecting and marking training data (more than 20 thousand images).









