A RealTime Small Target Vehicle Detection Algorithm with an Improved YOLOv11m Network Model

This project focuses on the development of a real-time small target vehicle detection algorithm using the YOLOv11m network model. The primary goal is to identify and classify various vehicle types from the VisDrone dataset, which includes categories such as awning-tricycle, bicycle, bus, car, motor, pedestrian, people, tricycle, truck, and van. YOLOv11m, a state-of-the-art deep learning model known for its high accuracy and fast inference speed, is utilized to detect small targets and vehicles in diverse environments. The model is trained to detect and classify these objects in real-time, ensuring efficient and accurate detection even in challenging scenarios with varying lighting conditions, occlusions, and cluttered backgrounds. The proposed algorithm aims to enhance automated surveillance systems, traffic monitoring, and autonomous vehicle navigation by providing robust detection of small and medium-sized vehicles. Through rigorous training on the VisDrone dataset, the algorithm demonstrates a promising balance of accuracy and speed, making it a suitable solution for practical applications in smart cities, transportation systems, and security surveillance. Keywords—Real-time detection, YOLOv11m, small target detection, vehicle detection, VisDrone dataset, deep learning, object classification, surveillance, autonomous vehicles.