Object Detection Algorithm for Autonomous Driving Scenes Based on Improved YOLOv8
A FAN-YOLOv8n autonomous driving detection algorithm is proposed to address the difficulties in detecting occluded and small targets in autonomous driving scenarios.Firstly,an enhanced feature field of view module(EFFVM)is designed to enhance the extraction of local features in the backbone of the model and improve its ability to detect oc-cluded targets.Secondly,a shallower feature layer P2 detection head has been added to the model head to improve the de-tection performance of the model for small targets.Then,a feature guidance module(FGM)is adopted at the neck of the model to fuse shallow and deep feature information,enabling better feature interaction between the two layers and making the model more focused on fine-grained features.Finally,a feature layer fusion module(FLFM)is proposed to fuse multi-scale feature layers and perform feature enhancement,enabling the model to adaptively detect targets of different scales.The experimental results show that on the SODA10M dataset and some BDD100K dataset,the improved model's mAP0.5 improves by 7 percentage points and 6.5 percentage points compared to the original YOLOv8n model,making it suitable for practical autonomous driving detection tasks.
万方数据
维普
