vehicle_counting_tensorflow

Why we included this project

Traffic-camera analytics usually settle for a count; this project also records what kind of vehicle passed, its color, size, direction of travel, and speed in km/h. Built on the TensorFlow Object Counting API, it detects and classifies cars, trucks, buses, bicycles, and motorcycles from video, crops each detected vehicle out of the frame, and writes the run to a CSV that covers type, color, direction, and speed. That is handy when prototyping parking-lot, garage, or campus traffic monitoring. Manage expectations: this is a sample reference implementation, not a maintained product, so plan on setup effort, model tuning, and adapting the code to your own footage. Use it as a documented starting point for seeing how detection, tracking, and speed measurement work together.