arxivcs.CV2026-06-26
Scene and Human in One World: Reconstruction in a Feedforward Pass
Boao Shi, Qiao Feng, Yiming Huang, Lingjie Liu
Reconstructing humans in dynamic scenes from moving monocular cameras remains challenging due to scale ambiguity, human-scene misalignment, and occlusion interference. Rather than treating human mesh recovery and scene reconstruction as separate tasks, we believe that accurate hu…