OpenCV as a traffic analysis tool
Can logistics benefit from machine learning? One field of application is certainly traffic. Software developer Andrey Nikishaev has created a series of algorithms that analyse traffic volume over a period of time for this very purpose. They use CCTV as the basis and generate automated measurement results through the use of OpenCV in four steps. First, they determine the appropriate picture section. After that, the code blurs at the edges, followed by the whole picture, based on a binary threshold. In the final step, a set ROI mask is laid over the constructed area. Now the volume of traffic can be analysed by counting the black and white pixels. The method is not fully precise – it currently has an accuracy level of 70-85 per cent. Nikishaev prefers to search for optimisations in an open source manner and you can review his code on GitHub.
We’d like to thank Andrey Nikishaev for his great work.
Read the entire article, Counting Road Traffic Capacity with OpenCV.