Research map / Detection and segmentation / Open-set and few-shot detection
What and Where the Themes Dominate in Image
Xinyu Xiao, Lingfeng Wang, Shiming Xiang, Chunhong Pan
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 concept (theme) attention to LSTM-ATT (ablation) | LSTM-ATT · COCO (MLE) | CIDEr: 106.5 -> 106.7 | Improved |
| add region attention to LSTM-ATT (ablation) | LSTM-ATT · COCO (MLE) | CIDEr: 106.5 -> 108.6 | Improved |
| combine concept and region attention (WWT) (ablation) | LSTM+RATT · COCO (MLE) | CIDEr: 108.6 -> 109.0 | Improved |
| add distinctive learning (ablation) | WWT · COCO (MLE) | CIDEr: 109.0 -> 110.9 | Improved |
| add dynamic reward weight (ablation) | WWT+D · COCO (MLE) | CIDEr: 110.9 -> 111.4 | Improved |
| add distinctive learning under RL training (ablation) | WWT · COCO (RL) | CIDEr: 116.4 -> 117.0 | Improved |
| add dynamic reward weight under RL training (ablation) | WWT+D · COCO (RL) | CIDEr: 117.0 -> 117.5 | Improved |
| add concept attention to LSTM-ATT under RL (ablation) | LSTM-ATT · COCO (RL) | CIDEr: 113.3 -> 114.5 | Improved |
| add region attention to LSTM-ATT under RL (ablation) | LSTM-ATT · COCO (RL) | CIDEr: 113.3 -> 115.5 | Improved |
| switch training from MLE to RL (CIDEr reward) | WWT+DW · COCO | CIDEr: 111.4 -> 117.5 | Improved |
| switch training from MLE to RL | WWT · Flickr30K | CIDEr: 73.5 -> 88.2 | Improved |
| switch training from MLE to RL | WWT · COCO test | CIDEr: 112.1 -> 118.8 | Improved |
Closest papers by meaning
- Paying Attention to Descriptions Generated by Image Captioning Models (ICCV 2017 · similarity 0.90)
- Aligning Where to See and What to Tell: Image Captioning with Region-Based Attention and Scene-Specific Contexts (TPAMI 2016 · similarity 0.89)
- Human Attention in Image Captioning: Dataset and Analysis (ICCV 2019 · similarity 0.87)
- Semantic and Expressive Variations in Image Captions Across Languages (CVPR 2025 · similarity 0.86)
- Enhancing Descriptive Captions with Visual Attributes for Multimodal Perception (CVPR 2026 · similarity 0.85)
- CapRL: Stimulating Dense Image Caption Capabilities via Reinforcement Learning (ICLR 2026 · 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.
