Vehicle Tracking and Counting for Intersection Analysis of Traffic Flow Movement

Urban transportation infrastructure is critical for city development, but the rapid increase in vehicle use in Indonesia has outpaced road expansion, leading to severe traffic congestion. This research aims to contribute to addressing traffic congestion by identifying traffic flow parameters within a traffic space and its repercussions on neighboring traffic flows using a vehicle tracking system to quantify vehicles and their directional movements. The research employs a methodology which includes dataset management, model training, vehicle tracking development using ByteTrack technology, and evaluation. The system outputs a processed video display and a JSON file containing vehicle movement counting data, achieving an 86.9% counting accuracy in tests with CCTV recordings from various days and times. While the results demonstrate high performance, factors such as camera angle, and bright sunlight as well as rainy weather conditions present challenges.