RayFusion: Ray Fusion Enhanced Collaborative Visual Perception

Tong Wang (Zhejiang University) · Shaohong Wang (Zhejiang University) · Lu Bin (Zhejiang University) · Xinyu Xiao (Alibaba Group) · Hanzhi Zhong (Zhejiang University) · Bowen Pang (Zhejiang University) · Zhiyu Xiang (Zhejiang University) · Hangguan Shan (Zhejiang University) · Eryun Liu (Zhejiang University)

3d object detection autonomous drivingcamera-based systemscollaborative visual perceptioncomprehensive experiments depth estimation detection performancefalse positive predictionsperformance advancementpublicly available coderay occupancy informationray-based fusionredundancy reductionsensor limitationsstate-of-the-art models

Collaborative visual perception methods have gained widespread attention in the autonomous driving community in recent years due to their ability to address sensor limitation problems. However, the absence of explicit depth information often makes it difficult for camera-based perception systems, e.g., 3D object detection, to generate accurate predictions. To alleviate the ambiguity in depth estimation, we propose RayFusion, a ray-based fusion method for collaborative visual perception. Using ray occupancy information from collaborators, RayFusion reduces redundancy and false positive predictions along camera rays, enhancing the detection performance of purely camera-based collaborative perception systems. Comprehensive experiments show that our method consistently outperforms existing state-of-the-art models, substantially advancing the performance of collaborative visual perception. Our code will be made publicly available.