YOLOv11-ASM: Enhancing Real-Time Object Detection for Autonomous Driving

This paper presents YOLOv11-ASM, a lightweight object detection model designed for autonomous driving, aiming to enhance detection accuracy and robustness in complex environments. The model integrates a Spatial-Aware Feature Modulation Network (SAFMN) to dynamically adjust feature representations via spatial context, improving adaptability. A Multi-Level Cross Attention (MLCA) mechanism is introduced to effectively fuse multi-scale features, while an Adaptive Task-Focused Loss (ATFL) function is designed to optimize regression performance, particularly benefiting small object detection and mitigating class imbalance. Experimental evaluations on the COCO128 dataset demonstrate that YOLOv11-ASM achieves superior performance compared to baseline models, with a precision of 0.939, recall of 0.783, and mAP@50 and mAP@50:95 reaching 0.919 and 0.789, respectively. The proposed model improves mAP@50 by 3.5%–5.9% over existing approaches, showcasing its strong generalization and real-time detection capabilities. These results highlight the model's potential for deployment in real-world autonomous systems requiring accurate and efficient object detection.