MambaOut: Do We Really Need Mamba for Vision?*

Weihao Yu, Xinchao Wang

arXiv:2405.07992 · 2026-07-27 공개 · arXiv · PDF

object-detection image-classification attention-mechanism segmentation mamba state-space-model long-sequence ssm

Abstract

Mamba, an architecture with RNN-like token mixer of state space model (SSM), was recently introduced to address the quadratic complexity of the attention mechanism and subsequently applied to vision tasks. Nevertheless, the performance of Mamba for vision is often underwhelming when compared with convolutional and attention-based models. In this paper, we delve into the essence of Mamba, and conceptually conclude that Mamba is ideally suited for tasks with long-sequence and autoregressive characteristics. For vision tasks, as image classification on ImageNet does not align with either characteristic, we hypothesize that Mamba is not necessary for this task; Detection and segmentation tasks on COCO or ADE20K are also not autoregressive, yet they adhere to the long-sequence characteristic, so we believe it is still worthwhile to explore Mamba’s potential for these tasks. To empirically verify our hypotheses, we construct a series of models named MambaOut through stacking Mamba blocks while removing their core token mixer, SSM. Experimental results strongly support our hypotheses. Specifically, our MambaOut model surpasses all visual Mamba models on ImageNet image classification, indicating that Mamba is indeed unnecessary for this task. As for detection and segmentation, MambaOut cannot match the performance of state-of-the-art visual Mamba models, demonstrating the potential of Mamba for long-sequence visual tasks.

한국어 요약

한 줄 요약

이미지넷 분류에서는 Mamba가 불필요하지만, 감지 및 세그멘테이션에서는 잠재력이 있다.

핵심 기여도

핵심 아이디어

Mamba는 RNN과 유사한 **structured state space model (SSM)**을 기반으로 하며, **long-sequence** 및 **autoregressive** 작업에 적합하다는 점에서 기존 Transformer의 제한을 극복할 수 있다는 기대가 있었다. 그러나 시각 작업은 대부분 **비자연적**(non-autoregressive)이며, 전체 이미지를 한 번에 볼 수 있어 **causal mode**가 불필요하다. 따라서 Mamba의 핵심인 SSM은 시각 작업에서 오히려 성능 저하 요인이 될 수 있다. 이에 따라, 저자는 **Gated CNN** 기반의 **MambaOut** 모델을 제안하며, SSM을 제거한 형태로 Mamba의 필요성을 검증한다.

기술적 접근법

주요 결과

의의 및 한계

실용적 활용