ZipMap：基于测试时训练的线性时间有状态三维重建
ZipMap: Linear-Time Stateful 3D Reconstruction via Test-Time Training
用测试时训练层把整组图像压缩成紧凑的隐场景状态，将多视图前馈重建的二次复杂度降到线性，视觉研究者可直接借鉴这一「有状态前馈」范式处理大规模图像集与实时场景查询。
原文依据 ZipMap employs test-time training layers to zip an entire image collection into a compact hidden scene state in a single forward pass
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论前馈 Transformer 模型推动了三维视觉的快速进展，但 VGGT 与 π³ 等最优方法的计算开销随输入图像数量呈二次增长，使其在大规模图像集合上效率低下。序列式重建方法虽降低了这一开销，却牺牲了重建质量。我们提出 ZipMap，一种有状态的前馈模型，在达到线性时间、双向三维重建的同时，精度与二次时间方法持平或更优。ZipMap 采用测试时训练层，在单次前向传播中把整个图像集合压缩为紧凑的隐场景状态，从而可在单块 H100 GPU 上于 10 秒内重建超过 700 帧，比 VGGT 等最优方法快 20 倍以上。此外，我们展示了有状态表征在实时场景状态查询中的优势，及其向序列式流式重建的扩展。
英文原文摘要 996 字符
Feed-forward transformer models have driven rapid progress in 3D vision, but state-of-the-art methods such as VGGT and \pi^3 have a computational cost that scales quadratically with the number of input images, making them inefficient when applied to large image collections. Sequential-reconstruction approaches reduce this cost but sacrifice reconstruction quality. We introduce ZipMap, a stateful feed-forward model that achieves linear-time, bidirectional 3D reconstruction while matching or surpassing the accuracy of quadratic-time methods. ZipMap employs test-time training layers to zip an entire image collection into a compact hidden scene state in a single forward pass, enabling reconstruction of over 700 frames in under 10 seconds on a single H100 GPU, more than 20x faster than state-of-the-art methods such as VGGT. Moreover, we demonstrate the benefits of having a stateful representation in real-time scene-state querying and its extension to sequential streaming reconstruction.