Multi-object tracking and detection system based on feature detection of the intelligent transportation system

This study aims to develop a system to detect traffic conditions by using computer vision. The system detects the presence of cars, motorcycles and pedestrians in the traffic; and also calculates the detected objects. The system detects the objects by using feature extraction method that is Histogram Oriented Gradient (HOG) and using linear Support Vector Machine (SVM) classifier. The system calculates the number of detected object by using the Kanade-Lucas-Tomasi (KLT) feature tracker. The implemented system has an average accuracy of 95.15%. Performance of HOG method and KLT algorithm was good enough to deal with the change of the brightness changes, but was not good enough to deal with pepper noise.