VMamba：视觉状态空间模型
VMamba: Visual State Space Model
AI总结：
本文提出 VMamba，将 Mamba 状态空间模型改造为线性时间复杂度的视觉骨干，通过二维选择性扫描模块高效建模视觉上下文，并在多种视觉任务上验证了其性能和扩展效率。
摘要：
在计算机视觉领域，设计计算高效的网络架构仍然是一项持续的需求。本文中，我们将状态空间语言模型 Mamba 改造为 VMamba，一种具有线性时间复杂度的视觉骨干网络。VMamba 的核心是一系列堆叠的视觉状态空间（VSS）模块，其中包含二维选择性扫描（SS2D）模块。SS2D 通过沿四条扫描路径遍历，弥合了一维选择性扫描的有序特性与二维视觉数据的非顺序结构之间的差距，从而有助于从不同来源和视角收集上下文信息。基于 VSS 模块，我们开发了一系列 VMamba 架构，并通过一系列架构和实现上的改进对其进行了加速。大量实验表明，VMamba 在多种视觉感知任务上展现出良好的性能，凸显了其相较于现有基准模型更优越的输入扩展效率。源代码可在 https://github.com/MzeroMiko/VMamba 获取。
英文摘要：
Designing computationally efficient network architectures remains an ongoing necessity in computer vision. In this paper, we adapt Mamba, a state-space language model, into VMamba, a vision backbone with linear time complexity. At the core of VMamba is a stack of Visual State-Space (VSS) blocks with the 2D Selective Scan (SS2D) module. By traversing along four scanning routes, SS2D bridges the gap between the ordered nature of 1D selective scan and the non-sequential structure of 2D vision data, which facilitates the collection of contextual information from various sources and perspectives. Based on the VSS blocks, we develop a family of VMamba architectures and accelerate them through a succession of architectural and implementation enhancements. Extensive experiments demonstrate VMamba's promising performance across diverse visual perception tasks, highlighting its superior input scaling efficiency compared to existing benchmark models. Source code is available at https://github.com/MzeroMiko/VMamba.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
