Salvato in:
Dettagli Bibliografici
Autori principali: Baty, Enis, Díaz, Alejandro Hernández, Davidson, Rebecca, Bridges, Chris, Hadfield, Simon
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2412.16146
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917351714193408
author Baty, Enis
Díaz, Alejandro Hernández
Davidson, Rebecca
Bridges, Chris
Hadfield, Simon
author_facet Baty, Enis
Díaz, Alejandro Hernández
Davidson, Rebecca
Bridges, Chris
Hadfield, Simon
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2412_16146
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Mamba2D: A Natively Multi-Dimensional State-Space Model for Vision Tasks
Baty, Enis
Díaz, Alejandro Hernández
Davidson, Rebecca
Bridges, Chris
Hadfield, Simon
Computer Vision and Pattern Recognition
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.
title Mamba2D: A Natively Multi-Dimensional State-Space Model for Vision Tasks
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.16146