Enhancing sustainable development with an intelligent vehicle detection system for a green university

This study proposes a replicable framework for automated vehicle detection and CO₂ estimation using YOLOv8 and edge computing. Deployed at Buriram Rajabhat University, the system integrates virtual line-crossing with ByteTrack and Kalman filter stabilization to ensure counting reliability across existing CCTV infrastructure. Performance evaluation showed a 92.3% F1-score, establishing an empirical baseline for campus carbon footprint management and supporting SDGs 11 and 13 through low-cost, AI-driven monitoring. Integrates YOLOv8 detection with ByteTrack and Kalman filtering for robust real-time vehicle counting on existing surveillance infrastructure. Incorporates a Cold-start Correction Factor into the emission estimation formula to enhance the accuracy of intra-campus carbon footprint data. Provides a scalable, low-cost methodological pipeline for academic institutions to monitor sustainability indicators without specialized sensing hardware.