[论文] ZipDepth: Bringing Lightweight Zero-Shot Monocular Depth Anywhere, on ...

论文概要

研究领域: CV
作者: Fabio Tosi, Luca Bartolomei, Matteo Poggi
发布时间: 2026-07-10
arXiv: 2507.08183

中文摘要

单目深度估计通过基础模型实现了显著的零样本泛化进展，但其计算需求使嵌入式和移动平台望而却步。轻量级替代方案存在，但几乎仅在单域自监督范式下开发，在域迁移时悄然失效。本文提出ZipDepth，一种紧凑的单目深度网络，通过将高效的可重参数化编码器-解码器与来自基础模型的大规模知识蒸馏相结合，弥合了这一差距。ZipDepth仅含610万参数，从服务器GPU到功耗受限设备均能以实时速度运行，在五个基准测试上实现了轻量级模型中零样本精度与部署效率的最佳权衡，向参数量50倍于己的基础模型精度迈出了重要一步。

原文摘要

Monocular depth estimation has seen remarkable progress through foundation models achieving robust zero-shot generalization, yet their computational demands place them far beyond the reach of embedded and mobile platforms. Lightweight alternatives exist, but have been developed almost exclusively within single-domain, self-supervised paradigms, failing silently under domain shift. We present ZipDepth, a compact monocular depth network that bridges this gap by combining an efficient reparameterizable encoder-decoder with large-scale knowledge distillation from a foundation model over a large multi-domain training set. Comprising just 6.1M parameters, ZipDepth runs at real-time rates from server GPUs to power-constrained devices, achieving the best trade-off between zero-shot accuracy and de...

自动采集于 2026-07-11

#论文 #arXiv #CV #小凯