Research map / Vision-language and multimodal / Cross-modal learning
Weakly Aligned Cross-Modal Learning for Multispectral Pedestrian Detection
Lu Zhang, Xiangyu Zhu, Xiangyu Chen, Xu Yang, Zhen Lei, Zhiyong 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 Region Feature Alignment (RFA) module (ablation) | AR-CNN · KAIST shift robustness (S45) | std of MR^T over shifts: 9.77 -> 1.24 | Improved |
| add Region Feature Alignment (RFA) module (ablation) | AR-CNN · KAIST shift robustness (origin) | MR^T: 12.94 -> 10.90 | Improved |
| add RoI jitter strategy on top of RFA (ablation) | AR-CNN · KAIST shift robustness (origin) | MR^T: 10.90 -> 9.87 | Improved |
| add RoI jitter strategy on top of RFA (ablation) | AR-CNN · KAIST shift robustness (S45 std) | std of MR^T over shifts: 2.59 -> 1.71 | Improved |
| add confidence-aware fusion on top of RFA and RoI jitter (ablation) | AR-CNN · KAIST shift robustness (origin) | MR^T: 9.87 -> 8.26 | Improved |
| add confidence-aware fusion on top of RFA and RoI jitter (ablation) | AR-CNN · KAIST shift robustness (S45 std) | std of MR^T over shifts: 1.71 -> 1.24 | Improved |
| add RFA, RoI jitter and confidence-aware fusion (full AR-CNN) versus MSDS-RCNN | AR-CNN · KAIST-Paired all-day | MR^T: 12.51 -> 8.26 | Improved |
| add AR-CNN versus MSDS-RCNN on color annotation | AR-CNN · KAIST-Paired all-day | MR^C: 11.28 -> 9.86 | Improved |
| add AR-CNN versus best prior method on CVC-14 night | AR-CNN · CVC-14 night | MR: 30.8 -> 18.1 | Improved |
| shift color image by (-6,6) pixels at test time | Baseline detector · KAIST reasonable | MR: 15.18 -> 25.10 | Hurt |
| shift color image by (1,-1) pixels at test time | Baseline detector · KAIST reasonable | MR: 15.18 -> 14.68 | Improved |
Closest papers by meaning
- Learning Cross-Modal Deep Representations for Robust Pedestrian Detection (CVPR 2017 · similarity 0.91)
- Improving Multispectral Pedestrian Detection by Addressing Modality Imbalance Problems (ECCV 2020 · similarity 0.89)
- When Pedestrian Detection Meets Multi-Modal Learning: Generalist Model and Benchmark Dataset (ECCV 2024 · similarity 0.88)
- Robust Pedestrian Detection with Uncertain Modality (AAAI 2026 · similarity 0.87)
- Causal Mode Multiplexer: A Novel Framework for Unbiased Multispectral Pedestrian Detection (CVPR 2024 · similarity 0.86)
- Multispectral Pedestrian Detection with Sparsely Annotated Label (AAAI 2025 · similarity 0.84)
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.
