RefineAny3D: Depth Refinement as Semantic Alignment for Monocular 3D Detection

作者：Zhihao Zhang, Gengwei Zhang, Tianlong Chen, Xiaoming Liu · 单位：Michigan State University, University of North Carolina at Chapel Hill · 会议/期刊：arXiv preprint · 方向：cs.CV · 发布日期：2026-08-11