Application for Recognition of Road Transport Using TensorFlow

The use of intelligent transport systems gives us an opportunity to solve the issues of reducing traffic jams, improving road safety and increasing energy efficiency. There was developed a system within this study that detects and classifies the vehicles on a video obtained from the intersection surveillance cameras. The presented system uses the SORT algorithm to track objects on video and a neural network architecture Mask R-CNN to recognize and classify objects on the video frames. The developed system allows to classify cars into 10 categories. And the accuracy of objects recognition reached 79%.