CORE: Cooperative Reconstruction for Multi-Agent Perception

Binglu Wang, Lei Zhang, Zhaozhong Wang, Yongqiang Zhao, Tianfei Zhou; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023, pp. 8710-8720

This paper presents CORE, a conceptually simple, effective and communication-efficient model for multi-agent cooperative perception. It addresses the task from a novel perspective of cooperative reconstruction, based on two key insights: 1) cooperating agents together provide a more holistic observation of the environment, and 2) the holistic observation can serve as valuable supervision to explicitly guide the model learning how to reconstruct the ideal observation based on collaboration. CORE instantiates the idea with three major components: a compressor for each agent to create more compact feature representation for efficient broadcasting, a lightweight attentive collaboration component for cross-agent message aggregation, and a reconstruction module to reconstruct the observation based on aggregated feature representations. This learning-to-reconstruct idea is task-agnostic, and offers clear and reasonable supervision to inspire more effective collaboration, eventually promoting perception tasks. We validate CORE on two large-scale multi-agent percetion dataset, OPV2V and V2X-Sim, in two tasks, i.e., 3D object detection and semantic segmentation. Results demonstrate that CORE achieves state-of-the-art performance, and is more communication-efficient.