Lite Any Stereo V2：更快、更强大、更高效的零样本立体匹配技术
Lite Any Stereo V2: Faster and Stronger Efficient Zero-Shot Stereo Matching
摘要
近年来，立体匹配技术取得了显著进展，其精度也得到了提升。不过，这些方法通常依赖庞大的模型、大量的计算资源或额外的基础模型先验信息，因此难以在资源有限的平台上应用。相比之下，高效的立体匹配模型虽然推理速度更快，但往往难以实现强大的零样本泛化能力。本文通过引入Lite Any Stereo V2（LAS2）来挑战这一假设——这是一种超快速模型系列，专为高效零样本立体匹配设计。LAS2从架构和训练两个方面进行了优化。在架构上，我们重新设计了适用于实际部署环境的高效立体匹配模型，并提出了一种仅基于2D数据的成本聚合框架，该框架更注重实际推理延迟，而非单纯的理论MAC值。在训练方面，我们采用了三阶段策略，结合合成监督、自蒸馏以及现实世界知识的传递机制。为了提升现实场景中的伪监督的可靠性，我们还引入了伪标签过滤和误差限制机制，从而让合成数据与真实数据之间的转换更加平滑。LAS2被设计成一系列模型，包括适用于不同效率要求的前馈变体，以及能够实现更高精度的迭代变体。大量实验表明，LAS2在保持较低延迟的同时，实现了高效立体匹配方法中最高的精度。具体来说，LAS2-H在零样本性能上优于迭代式方法Fast-FoundationStereo，在H200和Orin设备上分别实现了1.8倍和2.7倍的推理速度提升。相关项目页面、演示代码及文档可访问https://tomtomtommi.github.io/LiteAnyStereoV2/。
English Abstract
Recent advances in stereo matching have achieved remarkable accuracy, but often rely on large models, heavy computation, or additional foundation-model priors, making them difficult to deploy on resource-constrained platforms. In contrast, efficient stereo models offer faster inference but are commonly considered less capable of strong zero-shot generalization. In this paper, we challenge this assumption by introducing Lite Any Stereo V2 (LAS2), an ultra-fast model series designed for efficient zero-shot stereo matching. LAS2 is developed from both architecture and training perspectives. Architecturally, we revisit efficient stereo design under practical deployment settings and propose a 2D-only cost aggregation framework, optimized for real inference latency rather than theoretical MACs alone. For training, we develop a three-stage strategy that combines synthetic supervision, self-distillation, and real-world knowledge distillation. To improve the reliability of real-world pseudo supervision, we further introduce pseudo-label filtering and an error-clamping operation, enabling smoother synthetic-to-real transfer. We instantiate LAS2 as a family of models, including feed-forward variants for different efficiency budgets and an iterative variant for higher accuracy. Extensive experiments show that LAS2 achieves state-of-the-art accuracy among efficient stereo methods while maintaining significantly lower latency. Specifically, LAS2-H achieves stronger overall zero-shot performance than the iterative method Fast-FoundationStereo, with 1.8x and 2.7x faster inference on H200 and Orin, respectively. The project page, demos, and code are available at https://tomtomtommi.github.io/LiteAnyStereoV2/.