Research map / Vision-language and multimodal / Cross-modal learning
Distractors-Immune Representation Learning with Cross-modal Contrastive Regularization for Change Captioning
Yunbin Tu, Liang Li, Li Su, Chenggang Yan, Qingming Huang
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 DIRL to vanilla Transformer (ablation) | Transformer change captioner · CLEVR-DC | CIDEr: 79.6 -> 81.8 | Improved |
| add CCR to vanilla Transformer (ablation) | Transformer change captioner · CLEVR-DC | CIDEr: 79.6 -> 82.7 | Improved |
| add both DIRL and CCR (ablation) | Transformer change captioner · CLEVR-DC | CIDEr: 79.6 -> 84.1 | Improved |
| add both DIRL and CCR (ablation) | Transformer change captioner · CLEVR-DC | SPICE: 15.7 -> 16.8 | Improved |
| use only one DIRL term (non-perturbation or discrimination) instead of both (ablation) | DIRL · CLEVR-DC | CIDEr: 79.6 -> 79.3 / 79.5 | No clear effect |
| combine both DIRL terms (correlate matching channels and decorrelate different ones) (ablation) | DIRL · CLEVR-DC | CIDEr: 79.3 / 79.5 -> 81.8 | Improved |
| DIRL+CCR vs prior SOTA SCORER+CBR | DIRL+CCR · CLEVR-DC | BLEU-4: 49.4 -> 51.4 | Improved |
| DIRL+CCR vs SCORER+CBR on METEOR | DIRL+CCR · CLEVR-DC | METEOR: 33.4 -> 32.3 | Hurt |
| DIRL+CCR vs VIXEN-CLIP pretrained on large editing data | DIRL+CCR · Image Editing Request | CIDEr: 38.1 -> 34.1 | Hurt |
| DIRL+CCR vs SMART on Spot-the-Diff | DIRL+CCR · Spot-the-Diff | SPICE: 19.0 -> 19.9 | Improved |
| DIRL+CCR vs SMART on CLEVR-Change | DIRL+CCR · CLEVR-Change | CIDEr: 123.6 -> 123.6 | No clear effect |
Closest papers by meaning
- DECIDER: Difference-aware Contrastive Diffusion Model with Adversarial Perturbations for Image Change Captioning (AAAI 2025 · similarity 0.90)
- Region-aware Difference Distilling with Attribute-guided Contrastive Regularization for Change Captioning (AAAI 2025 · similarity 0.88)
- VIXEN: Visual Text Comparison Network for Image Difference Captioning (AAAI 2024 · similarity 0.86)
- Connect, Collapse, Corrupt: Learning Cross-Modal Tasks with Uni-Modal Data (ICLR 2024 · similarity 0.86)
- On the Value of Cross-Modal Misalignment in Multimodal Representation Learning (NeurIPS 2025 · similarity 0.85)
- Revisiting Change Captioning from Self-supervised Global-Part Alignment (AAAI 2025 · similarity 0.85)
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.
