Efficient Collection and Management Technology of Spatiotemporal Data for Intelligent Connected Vehicles Based on Computer Vision

The burgeoning realm of intelligent connected vehicle technology necessitates advanced data acquisition and management solutions to handle vast amounts of spatiotemporal data efficiently. This paper introduces a novel method using the YOLOv5 algorithm, enhancing data collection speed and accuracy through optimized computer vision techniques. Our approach leverages real-time data acquisition, sophisticated data feature processing, and enhanced spatiotemporal data storage and indexing. Experimentally, we demonstrated substantial improvements in data processing: data acquisition speeds increased linearly with frame rates, reaching up to 120 fps, and storage efficiency in Apache Cassandra outperformed MySQL, showcasing higher data throughput especially in large-scale datasets. The findings confirm that integrating YOLOv5 and multi-sensor fusion significantly boosts the environmental perception capabilities of connected vehicles, making this approach a promising avenue for future enhancements in intelligent transportation systems. The research underscores the critical role of efficient data management technologies in advancing the functionality and safety of smart vehicles.