Marigold V2: Revisiting Diffusion Transformers for Monocular Depth Estimation

• 类型：arxiv
• 标识：2609.08084
• 链接：https://arxiv.org/abs/2609.08084
• 主分类：multimodal
• 形态：application
• TLDR：Monocular depth estimation is a ubiquitous yet highly ill-posed computer vision task, with downstream applications in scene reconstruction, computational photography, and robotics, among others. Despite the field's maturity, recent models still struggle to generalize to out-of-distribution inputs and to produce sharp and detailed depth maps. In this paper, we revisit Marigold, a set of techniques for repurposing modern image generation and editing models, powered by the diffusion transformer (DiT) architecture, into state-of-the-art monocular depth estimators. Our recipes target single-step in
• 待LLM分类：否
• 标题中文：Marigold V2：重访用于单目深度估计的扩散 Transformer
• TLDR中文：单目深度估计是一项普遍存在但高度不适定的计算机视觉任务，下游应用涵盖场景重建、计算摄影与机器人等。尽管该领域已较成熟，近期模型在分布外输入的泛化以及生成清晰细节深度图方面仍存在困难。本文重访 Marigold——一套利用基于扩散 Transformer（DiT）架构的现代图像生成与编辑模型改造为 SOTA 单目深度估计器的技术方案。我们的方法针对单步推理……
• 来源文件：
• /inbox/tom/_candidates/2026-09-09-agent-rag-longcontext-candidates.json