提出二维状态空间模型，让视觉任务更高效准确
Mamba2D: A Natively Multi-Dimensional State-Space Model for Vision Tasks
- 从零重构多维状态空间模型，原生处理图像二维结构
- 2700万参数达ImageNet 84.0%准确率，超越所有同类模型
- 适合追求高效高精度视觉模型的研究者与工程师
状态空间模型（SSMs）已成为Transformer的高效替代方案，但现有视觉SSM仍受自然语言处理背景的深层偏见影响。本文提出M2D-SSM，从零开始重新推导适用于多维数据的选择性状态空间技术。不同于将1D SSM任意按行扫描图像的做法，我们的M2D-SSM采用单次二维扫描，原生建模空间维度。在ImageNet-1K分类任务中，M2D-T仅用2700万参数即达到84.0%的top-1准确率，超越所有同规模的SSM类视觉模型；M2D-S进一步实现85.3%，创下基于SSM架构的新纪录。在下游任务中，Mamba2D在MS-COCO目标检测（3×调度）上取得52.2 box AP，ADE20K语义分割上达51.7 mIoU，展现出强大的泛化能力与规模化效率。代码已开源：https://github.com/cocoalex00/Mamba2D。
原文摘要 · Abstract (English)
State-Space Models (SSMs) have emerged as an efficient alternative to transformers, yet existing visual SSMs retain deeply ingrained biases from their origins in natural language processing. In this paper, we address these limitations by introducing M2D-SSM, a ground-up re-derivation of selective state-space techniques for multidimensional data. Unlike prior works that apply 1D SSMs directly to images through arbitrary rasterised scanning, our M2D-SSM employs a single 2D scan that factors in both spatial dimensions natively. On ImageNet-1K classification, M2D-T achieves 84.0% top-1 accuracy with only 27M parameters, surpassing all prior SSM-based vision models at that size. M2D-S further achieves 85.3%, establishing state-of-the-art results among SSM-based architectures. Across downstream tasks, Mamba2D achieves 52.2 box AP on MS-COCO object detection (3$\times$ schedule) and 51.7 mIoU on ADE20K segmentation, demonstrating strong generalisation and efficiency at scale. Source code is available at https://github.com/cocoalex00/Mamba2D.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。