WaveMamba: Wavelet-Driven Mamba Fusion for RGB-Infrared Object Detection
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911074501001216 |
|---|---|
| author | Zhu, Haodong Dong, Wenhao Yang, Linlin Li, Hong Yang, Yuguang Ren, Yangyang Zhu, Qingcheng Feng, Zichao Li, Changbai Lin, Shaohui Wang, Runqi Luo, Xiaoyan Zhang, Baochang |
| author_facet | Zhu, Haodong Dong, Wenhao Yang, Linlin Li, Hong Yang, Yuguang Ren, Yangyang Zhu, Qingcheng Feng, Zichao Li, Changbai Lin, Shaohui Wang, Runqi Luo, Xiaoyan Zhang, Baochang |
| contents | Leveraging the complementary characteristics of visible (RGB) and infrared (IR) imagery offers significant potential for improving object detection. In this paper, we propose WaveMamba, a cross-modality fusion method that efficiently integrates the unique and complementary frequency features of RGB and IR decomposed by Discrete Wavelet Transform (DWT). An improved detection head incorporating the Inverse Discrete Wavelet Transform (IDWT) is also proposed to reduce information loss and produce the final detection results. The core of our approach is the introduction of WaveMamba Fusion Block (WMFB), which facilitates comprehensive fusion across low-/high-frequency sub-bands. Within WMFB, the Low-frequency Mamba Fusion Block (LMFB), built upon the Mamba framework, first performs initial low-frequency feature fusion with channel swapping, followed by deep fusion with an advanced gated attention mechanism for enhanced integration. High-frequency features are enhanced using a strategy that applies an ``absolute maximum" fusion approach. These advancements lead to significant performance gains, with our method surpassing state-of-the-art approaches and achieving average mAP improvements of 4.5% on four benchmarks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_18173 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | WaveMamba: Wavelet-Driven Mamba Fusion for RGB-Infrared Object Detection Zhu, Haodong Dong, Wenhao Yang, Linlin Li, Hong Yang, Yuguang Ren, Yangyang Zhu, Qingcheng Feng, Zichao Li, Changbai Lin, Shaohui Wang, Runqi Luo, Xiaoyan Zhang, Baochang Computer Vision and Pattern Recognition Multimedia Leveraging the complementary characteristics of visible (RGB) and infrared (IR) imagery offers significant potential for improving object detection. In this paper, we propose WaveMamba, a cross-modality fusion method that efficiently integrates the unique and complementary frequency features of RGB and IR decomposed by Discrete Wavelet Transform (DWT). An improved detection head incorporating the Inverse Discrete Wavelet Transform (IDWT) is also proposed to reduce information loss and produce the final detection results. The core of our approach is the introduction of WaveMamba Fusion Block (WMFB), which facilitates comprehensive fusion across low-/high-frequency sub-bands. Within WMFB, the Low-frequency Mamba Fusion Block (LMFB), built upon the Mamba framework, first performs initial low-frequency feature fusion with channel swapping, followed by deep fusion with an advanced gated attention mechanism for enhanced integration. High-frequency features are enhanced using a strategy that applies an ``absolute maximum" fusion approach. These advancements lead to significant performance gains, with our method surpassing state-of-the-art approaches and achieving average mAP improvements of 4.5% on four benchmarks. |
| title | WaveMamba: Wavelet-Driven Mamba Fusion for RGB-Infrared Object Detection |
| topic | Computer Vision and Pattern Recognition Multimedia |
| url | https://arxiv.org/abs/2507.18173 |