Pith. sign in

REVIEW 12 cited by

MambaOut: Do We Really Need Mamba for Vision?

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2405.07992 v3 pith:4KYARUON submitted 2024-05-13 cs.CV cs.AIcs.LG

Weihao Yu , Xinchao Wang This is my paper · ORCID
classification cs.CVcs.AIcs.LG
keywords mambatasksmambaoutmodelsvisionlong-sequencevisualautoregressive
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original 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 does not align with either characteristic, we hypothesize that Mamba is not necessary for this task; Detection and segmentation tasks 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. The code is available at https://github.com/yuweihao/MambaOut

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 12 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. A Spatial-Spectral-Frequency Interactive Network for Multimodal Remote Sensing Classification

    cs.CV 2025-10 unverdicted novelty 6.0 of 10

    S2Fin uses sparse spatial-spectral attention and two-level frequency enhancement to improve few-sample multimodal remote-sensing classification by 0.4–2.7% overall accuracy over recent state-of-the-art.

  2. After the Party: Navigating the Mapping From Color to Ambient Lighting

    cs.CV 2025-08 unverdicted novelty 6.0 of 10

    A new paired dataset and Retinex-based network, RLN2, for restoring images captured under multiple colored light sources to ambient-normalized versions.

  3. UAVScenes: A Multi-Modal Dataset for UAVs

    cs.CV 2025-07 conditional novelty 6.0 of 10

    UAVScenes adds frame-wise image and LiDAR semantic labels, reconstructed 6-DoF poses, and 3D maps to 120k frames of the MARS-LVIG dataset, with six benchmark tasks.

  4. MUG: Pseudo Labeling Augmented Audio-Visual Mamba Network for Audio-Visual Video Parsing

    cs.CV 2025-07 conditional novelty 6.0 of 10

    MUG combines manually corrected pseudo-labels, cross-modal random track recombination, and a Mamba-Transformer network to reach new state-of-the-art F1 scores on the LLP audio-visual video parsing benchmark.

  5. StableCodec: Taming One-Step Diffusion for Extreme Image Compression

    eess.IV 2025-06 conditional novelty 6.0 of 10

    A one-step diffusion codec that compresses noisy latents at 64x and decodes with a single denoising step, setting state-of-the-art FID, KID, and DISTS at ultra-low bitrates.

  6. MambaHash: Visual State Space Deep Hashing Model for Large-Scale Image Retrieval

    cs.CV 2025-06 conditional novelty 5.0 of 10

    MambaHash uses a grouped, multi-directional Mamba backbone for deep supervised hashing and reports the highest mean average precision on CIFAR-10 and IMAGENET, with marginal gains on NUS-WIDE.

  7. Semi-Supervised State-Space Model with Dynamic Stacking Filter for Real-World Video Deraining

    cs.CV 2025-05 conditional novelty 5.0 of 10

    The paper introduces a dual-branch state-space video deraining model with a dynamic stacking filter and semi-supervised median stacking loss, showing top PSNR across three benchmarks and new downstream task gains on a...

  8. Selective Structured State Space for Multispectral-fused Small Target Detection

    cs.CV 2025-05 conditional novelty 5.0 of 10

    A Mamba-based multispectral detector with three new modules reports state-of-the-art accuracy on VEDAI at real-time speed and with 17 MB size, though it trails some methods on larger objects.

  9. Is Long Range Sequential Modeling Necessary For Colorectal Tumor Segmentation?

    cs.CV 2025-02 conditional novelty 5.0 of 10

    A convolutional UNet without long-range sequence modeling beat Transformer and Mamba models on a new 204-patient colorectal tumor CT dataset.

  10. VCMamba: Bridging Convolutions with Multi-Directional Mamba for Efficient Visual Representation

    cs.CV 2025-09 conditional novelty 4.0 of 10

    VCMamba reports that using convolutional feed-forward blocks for the first three stages followed by multi-directional Mamba blocks in the final stage yields 82.6% ImageNet-1K and 47.1 ADE20K mIoU at 31.5M parameters, ...

  11. Modernizing CNN-based Weather Forecast Model towards Higher Computational Efficiency

    cs.CV 2025-07 conditional novelty 4.0 of 10

    A 7-million-parameter convolutional weather model trains in 12 hours on one GPU and is reported to match or beat much larger AI and numerical weather models in medium-range forecasts.

  12. Hypergraph Mamba for Efficient Whole Slide Image Understanding

    cs.CV 2025-05 conditional novelty 4.0 of 10

    WSI-HGMamba integrates hypergraph convolution with bidirectional Mamba sequence modeling, reporting accurate whole-slide classification at lower FLOPs than transformer baselines.

Pith tools