English

MobileMamba: Lightweight Multi-Receptive Visual Mamba Network

Computer Vision and Pattern Recognition 2024-11-26 v1

Abstract

Previous research on lightweight models has primarily focused on CNNs and Transformer-based designs. CNNs, with their local receptive fields, struggle to capture long-range dependencies, while Transformers, despite their global modeling capabilities, are limited by quadratic computational complexity in high-resolution scenarios. Recently, state-space models have gained popularity in the visual domain due to their linear computational complexity. Despite their low FLOPs, current lightweight Mamba-based models exhibit suboptimal throughput. In this work, we propose the MobileMamba framework, which balances efficiency and performance. We design a three-stage network to enhance inference speed significantly. At a fine-grained level, we introduce the Multi-Receptive Field Feature Interaction(MRFFI) module, comprising the Long-Range Wavelet Transform-Enhanced Mamba(WTE-Mamba), Efficient Multi-Kernel Depthwise Convolution(MK-DeConv), and Eliminate Redundant Identity components. This module integrates multi-receptive field information and enhances high-frequency detail extraction. Additionally, we employ training and testing strategies to further improve performance and efficiency. MobileMamba achieves up to 83.6% on Top-1, surpassing existing state-of-the-art methods which is maximum x21 faster than LocalVim on GPU. Extensive experiments on high-resolution downstream tasks demonstrate that MobileMamba surpasses current efficient models, achieving an optimal balance between speed and accuracy.

Keywords

Cite

@article{arxiv.2411.15941,
  title  = {MobileMamba: Lightweight Multi-Receptive Visual Mamba Network},
  author = {Haoyang He and Jiangning Zhang and Yuxuan Cai and Hongxu Chen and Xiaobin Hu and Zhenye Gan and Yabiao Wang and Chengjie Wang and Yunsheng Wu and Lei Xie},
  journal= {arXiv preprint arXiv:2411.15941},
  year   = {2024}
}

Comments

14 pages

R2 v1 2026-06-28T20:10:39.261Z