Advancing vehicle detection for autonomous driving: integrating computer vision and machine learning techniques for real-world deployment

Road-object detection and recognition are crucial for self-driving vehicles to achieve autonomy. Detecting and tracking other vehicles is a key task, but deep-learning methods, while effective, demand high computational power and expensive hardware. This paper proposes a lightweight vehicle detection technique (LWVDT) designed for low-cost CPUs without compromising robustness, speed, or accuracy. Suitable for advanced driving assistance systems (ADAS) and autonomous vehicle subsystems, LWVDT combines computer vision techniques like color spatial feature extraction and Histogram of Oriented Gradients (HOG) with machine learning methods such as support vector machines (SVM) to optimize performance. The algorithm processes raw RGB images to generate vehicle boundary boxes and tracks them across frames. Evaluated using real-road images, videos, and the KITTI database under various conditions, LWVDT achieves up to 87% accuracy, demonstrating its effectiveness in diverse environments.