Depth Any Endoscopy：面向单目内窥镜自监督可泛化深度估计
Depth Any Endoscopy: Towards Self-Supervised Generalizable Depth Estimation in Monocular Endoscopy
把视觉基础模型通过双层 MoE（内部 LoRA/Adapter、外部域专家）适配到内窥镜深度估计，并用可学习梯度协调平衡深度与位姿网络优化，可直接迁移到其他跨域自监督单目深度任务。
原文依据 we propose Depth Any Endoscopy (DAE), a novel self-supervised framework for generalizable depth estimation in monocular endoscopy
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论单目深度估计是内窥镜三维重建与定位等应用中的核心技术。然而，现有方法大多主要关注域内深度估计，由于深度分布、光照条件和纹理模式的差异，这限制了它们的鲁棒性，使其无法在跨域场景中取得令人满意的表现。在本工作中，我们提出 Depth Any Endoscopy（DAE），一种用于单目内窥镜可泛化深度估计的新型自监督框架。具体而言，我们提出一种双层混合专家（MoE）适配范式，能够有效地将视觉基础模型定制到腹腔镜、结肠镜等多种内窥镜手术场景，以应对环境变化带来的挑战。在内部，我们在 MoE 架构中集成低秩适配与适配器模块，使模型能够灵活适应输入数据的特性。在外部，域特定专家的混合为训练稳定性提供定制化引导。此外，我们引入一种可学习的梯度协调机制，以动态平衡深度网络与位姿网络之间的优化，并提出一个语义分布校准模块，以增强深度预测的语义一致性。大量实验表明，所提出的 DAE 在零样本和域内深度估计场景中均达到最优性能。源代码已公开于 https://github.com/ShuweiShao/DAE。
英文原文摘要 1563 字符
Monocular depth estimation serves as a core technique in endoscopic applications such as 3D reconstruction and localization. However, most existing methods focus primarily on in-domain depth estimation, which limits their robustness and prevents them from delivering impressive cross-domain performance, due to variations in depth distributions, illumination conditions, and texture patterns. In this work, we propose Depth Any Endoscopy (DAE), a novel self-supervised framework for generalizable depth estimation in monocular endoscopy. To specify, we develop a dual-level Mixture-of-Experts (MoE) adaptation paradigm that effectively tailors Vision Foundation Models to diverse endoscopic procedures, such as laparoscopy and colonoscopy, accounting for the challenges posed by varying environments. Internally, we integrate LoRA and Adapter modules within the MoE architecture, allowing the model to flexibly adapt to the characteristics of input data. Externally, a mixture of domain-specific experts provides customized guidance to enhance the training stability. In addition, we introduce a learnable gradient harmonization mechanism to dynamically balance the optimization between the depth and pose networks, along with a semantic distribution calibration module that strengthens the semantic consistency of depth predictions. Extensive experiments demonstrate that the proposed DAE achieves state-of-the-art performance in both zero-shot and in-domain depth estimation scenarios. The source code is publicly available at https://github.com/ShuweiShao/DAE.