Research map / Vision-language and multimodal / Human and embodied modeling
Aligning Where to See and What to Tell: Image Captioning with Region-Based Attention and Scene-Specific Contexts
Kun Fu, Junqi Jin, Runpeng Cui, Fei Sha, Changshui Zhang
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-based attention (ablation) | our-base-greedy · MSCOCO | CIDEr-D: 70.7 -> 75.1 | Improved |
| add scene-factorized LSTMs (ablation) | our-base-greedy · MSCOCO | CIDEr-D: 70.7 -> 74.3 | Improved |
| add both region attention and scene factorization (ablation) | our-base-greedy · MSCOCO | CIDEr-D: 70.7 -> 78.3 | Improved |
| add scene factorization to region attention (ablation) | our-ra-greedy · MSCOCO | CIDEr-D: 75.1 -> 78.3 | Improved |
| add region attention to scene factorization (ablation) | our-sf-greedy · MSCOCO | CIDEr-D: 74.3 -> 78.3 | Improved |
| use beam search (beam 10) instead of greedy | our-(ra+sf) · MSCOCO | CIDEr-D: 78.3 -> 83.8 | Improved |
| use beam search (beam 10) instead of greedy | our-(ra+sf) · MSCOCO | BLEU-4: 24.6 -> 28.2 | Improved |
| add region attention to base (ablation) | our-base-greedy · MSCOCO | METEOR: 20.0 -> 22.2 | Improved |
| add region attention instead of scene factorization (BLEU-4) (ablation) | our-ra-greedy · MSCOCO | BLEU-4: 24.2 -> 23.5 | Hurt |
| add region attention to base (ablation) | our-base-greedy · MSCOCO | BLEU-4: 22.6 -> 23.5 | Improved |
| distort scene vector to another topic | our model · MSCOCO qualitative | caption content | Improved |
| use our-ra+sf-beam over mRNN for caption to image retrieval | our-ra+sf-beam · MSCOCO retrieval | R@1: 29.0 -> 29.3 | Improved |
Closest papers by meaning
- From Show to Tell: A Survey on Deep Learning-Based Image Captioning (TPAMI 2022 · similarity 0.91)
- Text-Guided Attention Model for Image Captioning (AAAI 2017 · similarity 0.90)
- Deep Visual-Semantic Alignments for Generating Image Descriptions (TPAMI 2016 · similarity 0.90)
- A Hierarchical Approach for Generating Descriptive Image Paragraphs (CVPR 2017 · similarity 0.90)
- Patch Matters: Training-free Fine-grained Image Caption Enhancement via Local Perception (CVPR 2025 · similarity 0.89)
- Aligning Forest and Trees in Images & Long Captions for Visually Grounded Understanding (NeurIPS 2026 · similarity 0.88)
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.
