Research map / Detection and segmentation / CNN features
DETR is Secretly a Multispectral Detector: Zero-Parameter Adaptation via Semantic Alignment
Xiangyang Li, Zhiwei Jiang, Wushuai Jin, Pengyang Niu, Chunna Tian, Lingqiao Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add asymmetric input organization (AIO) (ablation) | RT-DETR, query IR memory RGB · FLIR | mAP (%): 33.88 -> 47.62 | Improved |
| add asymmetric input organization (AIO) (ablation) | RT-DETR, query RGB memory IR · FLIR | mAP (%): 45.19 -> 49.21 | Improved |
| add spatial semantic alignment (ablation) | AIO, query IR memory RGB · FLIR | mAP (%): 47.62 -> 47.95 | Improved |
| add channel semantic alignment (ablation) | AIO, query IR memory RGB · FLIR | mAP (%): 47.62 -> 47.80 | Improved |
| add both spatial and channel alignment (ablation) | AIO, query IR memory RGB · FLIR | mAP (%): 47.62 -> 48.16 | Improved |
| add spatial semantic alignment (ablation) | AIO, query RGB memory IR · FLIR | mAP (%): 49.21 -> 49.48 | Improved |
| add channel semantic alignment (ablation) | AIO, query RGB memory IR · FLIR | mAP (%): 49.21 -> 49.58 | Improved |
| add both spatial and channel alignment (ablation) | AIO, query RGB memory IR · FLIR | mAP (%): 49.21 -> 49.82 | Improved |
| use RGB as query and IR as memory instead of reverse | ZPA-MDETR · FLIR | mAP (%): 48.16 -> 49.82 | Improved |
| use IR as query and RGB as memory instead of reverse | ZPA-MDETR · M3FD | mAP (%): 48.39 -> 50.58 | Improved |
| use RGB as query and IR as memory instead of reverse | ZPA-MDETR · LLVIP | mAP (%): 64.88 -> 69.14 | Improved |
| use RGB query IR memory instead of reverse | ZPA-MDETR · RGBTDronePerson | mAP (%): 20.08 -> 21.01 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MS-DETR — MS-DETR: Efficient DETR Training with Mixed Supervision (CVPR 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Semantic-Aligned Matching for Enhanced DETR Convergence and Multi-Scale Feature Fusion (IJCV 2024 · similarity 0.88)
- Accelerating DETR Convergence via Semantic-Aligned Matching (CVPR 2022 · similarity 0.88)
- DAMSDet: Dynamic Adaptive Multispectral Detection Transformer with Competitive Query Selection and Adaptive Feature Fusion (ECCV 2024 · similarity 0.88)
- SMR-Net: Semantic-Guided Mutually Reinforcing Network for Cross-Modal Image Fusion and Salient Object Detection (AAAI 2025 · similarity 0.86)
- MI-DETR: An Object Detection Model with Multi-time Inquiries Mechanism (CVPR 2025 · similarity 0.86)
- DiMSOD: A Diffusion-Based Framework for Multi-Modal Salient Object Detection (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
