Research map / Vision-language and multimodal / Human and embodied modeling
Group-Based Distinctive Image Captioning with Memory Difference Encoding and Attention
Jiuniu Wang, Wenjia Xu, Qingzhong Wang, Antoni B. Chan
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 DifDisCap to Transformer | Transformer · MSCOCO test | DisWordRate (%): 16.8 -> 19.7 | Improved |
| add DifDisCap to Transformer | Transformer · MSCOCO test | CIDErBtw: 74.8 -> 70.7 | Improved |
| add DifDisCap to Transformer (accuracy cost) | Transformer · MSCOCO test | CIDEr: 111.7 -> 107.1 | Hurt |
| add DifDisCap to M2Transformer+SCST | M2Transformer+SCST · MSCOCO test | DisWordRate (%): 17.3 -> 18.7 | Improved |
| add DifDisCap to M2Transformer+SCST (accuracy cost) | M2Transformer+SCST · MSCOCO test | CIDEr: 128.9 -> 127.2 | Hurt |
| add image-group-based training (ablation) | M2Transformer · MSCOCO test | DisWordRate (%): 16.4 -> 16.9 | Improved |
| add weighted distinctive loss (WeiDisLoss) (ablation) | M2Transformer+SCST + ImageGroup · MSCOCO test | DisWordRate (%): 17.4 -> 18.3 | Improved |
| add group-based differential memory attention (GDMA) (ablation) | M2Transformer + ImageGroup + WeiDisLoss · MSCOCO test | DisWordRate (%): 17.3 -> 18.6 | Improved |
| add Indicated Training (ablation) | M2Transformer + GDMA · MSCOCO test | DisWordRate (%): 18.6 -> 18.8 | Improved |
| use RegionCLIP region features instead of Bottom-up (ablation) | M2Transformer+SCST+DifDisCap · MSCOCO test | DisWordRate (%): 18.7 -> 18.9 | Improved |
| use ViLBERT region features instead of Bottom-up (ablation) | M2Transformer+SCST+DifDisCap · MSCOCO test | DisWordRate (%): 18.7 -> 18.3 | Hurt |
| raise group size K from 1 to 5 (ablation) | M2Transformer+SCST+DifDisCap · MSCOCO test | DisWordRate (%): 17.6 -> 18.7 | Mixed |
Closest papers by meaning
- On Distinctive Image Captioning via Comparing and Reweighting (TPAMI 2022 · similarity 0.90)
- GroupCap: Group-Based Image Captioning With Structured Relevance and Diversity Constraints (CVPR 2018 · similarity 0.89)
- Context-Aware Group Captioning via Self-Attention and Contrastive Features (CVPR 2020 · similarity 0.89)
- Semantic and Expressive Variations in Image Captions Across Languages (CVPR 2025 · similarity 0.86)
- VisMin: Visual Minimal-Change Understanding (NeurIPS 2024 · similarity 0.85)
- CAPability: A Comprehensive Visual Caption Benchmark for Evaluating Both Correctness and Thoroughness (NeurIPS 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.
