Research map / Vision-language and multimodal / Cross-modal learning
Towards Robust Multi-Modal Semantic Segmentation with Teacher-Student Framework and Hybrid Prototype Distillation
Jiaqi Tan, Xu Zheng, Yang 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 Hybrid Prototype Distillation (HPD) to the missing-modality framework (ablation) | M-SegFormer · DeLiVER EMM | mIoU (%): 39.36 -> 49.10 | Improved |
| add teacher feedback from non-dominant modality on top of HPD (ablation) | M-SegFormer · DeLiVER EMM | mIoU (%): 49.10 -> 49.91 | Improved |
| apply HPD and feedback to CMNeXt backbone | CMNeXt · DeLiVER EMM | mIoU (%): 32.84 -> 47.04 | Improved |
| HPD (without feedback) on CMNeXt full-modality performance (ablation) | CMNeXt · DeLiVER full modality | mIoU (%): 60.12 -> 58.73 | Hurt |
| add feedback recovers full-modality performance (ablation) | CMNeXt · DeLiVER full modality | mIoU (%): 58.73 -> 59.87 | Improved |
| use IFV map for cross-modal distillation instead of prototype (ablation) | RobustSeg · DeLiVER (MiT-B0) | full-modality mIoU (%): 61.92 -> 53.38 | Hurt |
| use class-prototype for cross-modal distillation (ablation) | RobustSeg · DeLiVER (MiT-B0) | EMM mIoU (%): 39.36 -> 49.06 | Improved |
| combine IFV map and prototype without dominant selection (ablation) | RobustSeg · DeLiVER (MiT-B0) | EMM mIoU (%): 49.06 -> 48.63 | Hurt |
| add dominant-modality selection for IFV map distillation (ablation) | RobustSeg HPD · DeLiVER (MiT-B0) | EMM mIoU (%): 48.63 -> 49.10 | Improved |
| teacher feedback using student logits (CE + KL) instead of modality features (ablation) | RobustSeg feedback · DeLiVER (MiT-B0) | student EMM mIoU (%): 49.10 -> 48.77 | Hurt |
| freeze dominant encoders and fine-tune only non-dominant modality during feedback (ablation) | RobustSeg feedback · DeLiVER (MiT-B0) | student EMM mIoU (%): 49.13 -> 49.91 | Improved |
| RobustSeg vs AnySeg on three robustness benchmarks (mean) | RobustSeg · DeLiVER, MUSES, MCubeS | mean robustness mIoU (%): DeLiVER 41.46 -> 45.16 | 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.
- MAGIC — MAGIC: Multimodal relAtional Graph adversarIal inferenCe for Diverse and Unpaired Text-Based Image Captioning (AAAI 2022) · better on 4 of 4 shared results
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
- Learning Modality-agnostic Representation for Semantic Segmentation from Any Modalities (ECCV 2024 · similarity 0.90)
- OmniSegmentor: A Flexible Multi-Modal Learning Framework for Semantic Segmentation (NeurIPS 2025 · similarity 0.88)
- Improving Semi-Supervised Semantic Segmentation with Sliced-Wasserstein Feature Alignment and Uniformity (CVPR 2025 · similarity 0.87)
- Toward Modality Gap: Vision Prototype Learning for Weakly-supervised Semantic Segmentation with CLIP (AAAI 2025 · similarity 0.87)
- Keep the Balance: A Parameter-Efficient Symmetrical Framework for RGB+X Semantic Segmentation (CVPR 2025 · similarity 0.87)
- UniMRSeg: Unified Modality-Relax Segmentation via Hierarchical Self-Supervised Compensation (NeurIPS 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.
