Qiaoyu Zhu

SiZeUp: Fast 3D Proxy from Aerial Images via Depth Ordinal Loss

This work introduces an ordinal depth consistency loss that enforces agreement between the relative depth ordering of rendered proxies and depth priors predicted by a monocular depth model, and achieves a 23-52$\times$ speedup over state-of-the-art proxy reconstruction pipelines while maintaining comparable proxy-level...