Early Redundancy Assessment in Cooperative Perception Using a Modified YOLO Architecture

Abstract

Cooperative perception enables connected and autonomous vehicles to overcome the inherent limitations of individual sensors, such as occlusion and restricted field of view, by sharing perceived objects among vehicles. While this enhances situational awareness, it introduces significant redundancy in dense traffic, leading to unnecessary computation and communication overhead. Unlike existing methods that filter redundancy only after multi-modal processing or transmission, this paper proposes a novel redundancy-aware perception framework that integrates redundancy assessment directly into the detection stage. We extend the YOLOv11x architecture to ingest lightweight spatial masks (occupancy and inverse-distance) derived from shared data, enabling the model to simultaneously perform 2D object detection and binary redundancy classification in a single inference step. Experimental results on the V2X-Sim dataset demonstrate that the proposed method achieves an F1-score of 0.97 in redundancy classification. By identifying redundant objects early, the framework enables critical resource-saving strategies, such as context-aware downsampling and selective offloading, without compromising the detection of novel, safety-critical objects.