SAM 3D：把图像中的一切三维化
SAM 3D: 3Dfy Anything in Images
把单图物体重建当作生成任务，用「合成数据预训练 + 真实数据对齐」的多阶段训练突破 3D 数据瓶颈；对做单目重建、3D 资产生成或需要大规模带位姿/纹理 3D 标注数据的工作，这套数据管线与训练范式可直接复用。
原文依据 combines synthetic pretraining with real-world alignment, breaking the 3D "data barrier"
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论本文提出 SAM 3D，一种以视觉为条件的生成式三维物体重建模型，可从单张图像预测几何、纹理与布局。SAM 3D 在自然图像中表现优异——这类图像中遮挡与场景杂乱十分常见，来自上下文的视觉识别线索作用更大。为实现这一点，我们构建了人在回路与模型在回路的标注流程，用于标注物体形状、纹理与位姿，从而以空前的规模提供视觉条件化的三维重建数据。我们在此数据上采用现代的多阶段训练框架进行学习，将合成数据预训练与真实世界对齐相结合，突破了三维领域的「数据壁垒」。相较近期工作我们取得显著提升，在真实物体与场景上的人类偏好测试中胜率至少为 5:1。我们将公开代码与模型权重、在线演示，以及一个新的、具有挑战性的野外三维物体重建基准。
英文原文摘要 902 字符
We present SAM 3D, a generative model for visually grounded 3D object reconstruction, predicting geometry, texture, and layout from a single image. SAM 3D excels in natural images, where occlusion and scene clutter are common and visual recognition cues from context play a larger role. We achieve this with a human- and model-in-the-loop pipeline for annotating object shape, texture, and pose, providing visually grounded 3D reconstruction data at unprecedented scale. We learn from this data in a modern, multi-stage training framework that combines synthetic pretraining with real-world alignment, breaking the 3D "data barrier". We obtain significant gains over recent work, with at least a 5:1 win rate in human preference tests on real-world objects and scenes. We will release our code and model weights, an online demo, and a new challenging benchmark for in-the-wild 3D object reconstruction.