AI RESEARCH

DoReMi: Bridging 3D Domains via Topology-Aware Domain-Representation Mixture of Experts

arXiv CS.CV

ArXi:2511.11232v2 Announce Type: replace Constructing a unified 3D scene understanding model has long been hindered by the significant topological discrepancies across different sensor modalities. While applying the Mixture-of-Experts (MoE) architecture is an effective approach to achieving universal understanding, we observe that existing 3D MoE networks often suffer from semantics-driven routing bias.