Research map / Vision-language and multimodal / Human and embodied modeling
Explicit Image Caption Editing
Zhen Wang, Long Chen, Wenbo Ma, Guangxing Han, Yulei Niu, Jian Shao, Jun Xiao
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 |
|---|---|---|---|
| weighted XE loss lambda=1.5 instead of 1.0 (ablation) | TIger · COCO-EE test | CIDEr-D: 190.0 -> 194.8 | Improved |
| weighted XE loss lambda=1.5 instead of 1.0 (ablation) | TIger · Flickr30K-EE test | CIDEr-D: 137.0 -> 148.3 | Improved |
| increase XE weight to 2.0 from 1.5 (ablation) | TIger · COCO-EE test | CIDEr-D: 194.8 -> 193.9 | Hurt |
| add Tagger_add on top of Tagger_del (ablation) | TIger · COCO-EE test | CIDEr-D: 193.7 -> 194.8 | Improved |
| add Tagger_del on top of Tagger_add (ablation) | TIger · COCO-EE test | CIDEr-D: 191.2 -> 194.8 | Improved |
| add Tagger_add on top of Tagger_del (ablation) | TIger · Flickr30K-EE test | CIDEr-D: 144.9 -> 148.3 | Improved |
| add Tagger_del on top of Tagger_add (ablation) | TIger · Flickr30K-EE test | CIDEr-D: 140.9 -> 148.3 | Improved |
| add pretrained ViLBERT weights (ablation) | TIger · COCO-EE test | CIDEr-D: 178.1 -> 194.8 | Improved |
| add pretrained ViLBERT weights (ablation) | TIger · Flickr30K-EE test | CIDEr-D: 140.8 -> 148.3 | Improved |
| use 4 editing rounds instead of 1 (ablation) | TIger · COCO-EE test | CIDEr-D: 180.2 -> 194.8 | Improved |
| use 5 editing rounds instead of 4 (ablation) | TIger · COCO-EE test | CIDEr-D: 194.8 -> 194.6 | No clear effect |
| use 3 editing rounds instead of 1 (ablation) | TIger · Flickr30K-EE test | CIDEr-D: 142.8 -> 148.3 | Improved |
Closest papers by meaning
- An Item Is Worth a Prompt: Versatile Image Editing with Disentangled Control (AAAI 2025 · similarity 0.84)
- ImgEdit: A Unified Image Editing Dataset and Benchmark (NeurIPS 2025 · similarity 0.83)
- Revisit Large-Scale Image-Caption Data in Pre-training Multimodal Foundation Models (ICLR 2025 · similarity 0.83)
- Preserve or Modify? Context-Aware Evaluation for Balancing Preservation and Modification in Text-Guided Image Editing (CVPR 2025 · similarity 0.83)
- Patch Matters: Training-free Fine-grained Image Caption Enhancement via Local Perception (CVPR 2025 · similarity 0.83)
- PairEdit: Learning Semantic Variations for Exemplar-based Image Editing (NeurIPS 2025 · similarity 0.83)
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.
