Research on Computer Vehicle Inspection and Its Impact on Urban Traffic Intelligence

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

With the acceleration of urbanization, urban traffic faces multiple challenges such as congestion and safety, accompanied by an explosive growth of traffic data. Promoting the intelligent development of urban transportation is of vital theoretical and practical value for improving traffic efficiency and travel experience. This paper adopts a literature review methodology to outline the technological shift from traditional machine learning classifiers to deep learning paradigms in vehicle detection. Traditional vehicle detection relying on SVM and Decision Trees suffers from manual feature extraction and severe operational bloat, yielding sluggish processing speeds that fail modern real-time demands. Conversely, the deep learning paradigm, led by YOLO, treats detection as a single regression problem to drastically enhance speed and accuracy. For comprehensive tracking, integrating YOLO with ByteTrack ensures continuous vehicle identification and precise lane occupancy analysis, while combining it with LSTM networks enables predicting traffic flow two hours in advance for dynamic signal control. To overcome optical limitations in extreme environments, multi-modal sensor fusion (combining YOLO with LiDAR/RADAR) and decentralized edge computing are reshaping smart city infrastructure. This evolutionary trajectory shifts urban mobility toward predictive governance, highlighting future needs for lightweight edge optimization, all-weather fusion, and model interpretability.