Vision-Aided Channel Prediction Based on Image Segmentation at Street Intersection Scenarios

Intelligent vehicular communication with vehicle-road collaboration capability is a key technology enabled by 6G, and the integration of various visual sensors on vehicles and infrastructures plays a crucial role. Moreover, accurate channel prediction is foundational to realizing intelligent vehicular communication. Traditional methods are still limited by the inability to balance accuracy and complexity based on substantial spectrum resource consumption and highly refined description of environment. In this paper, we propose a vision-aided channel prediction model that leverages out-of-band visual information for precise prediction of typical channel characteristics, including path loss, Rice K-factor, and delay spread. We first conduct extensive vehicle-to-infrastructure measurements at urban intersections, collecting synchronized channel and visual data. After post-processing and filtering, segmented images of the target vehicle are generated using YOLOv8 network to construct an image-channel dataset. These segmented images are then used as inputs to a ResNet-34 model for supervised learning. Finally, four experiments are designed to evaluate the model, which are scenario self-validation, cross-validation, generalization to varying vehicle appearances, and performance under different segmentation and prediction networks. Experimental results confirm that our model, trained with instance-segmented images, achieves high prediction accuracy and strong generalization across scenarios and target users. The proposed approach offers a promising and practical solution for intelligent channel prediction in future vehicular communication systems.