Research map / Detection and segmentation / CNN features
Image Caption with Global-Local Attention
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 object-level local features with attention to global feature (ablation) | LSTM captioner · MS COCO | CIDEr: 76.9 -> 85.7 | Helped |
| use only local object features instead of global feature (ablation) | LSTM captioner · MS COCO | CIDEr: 76.9 -> 73.6 | Hurt |
| add linear embedding layer to reduce feature to 1000-d (ablation) | GloLocAtt · MS COCO | CIDEr: 85.7 -> 87.3 | Helped |
| add one dropout layer after second LSTM (ablation) | GloLocAttEmb · MS COCO | CIDEr: 87.3 -> 89.2 | Helped |
| add second dropout layer after first LSTM (ablation) | GloLocAttEmb+OneDrop · MS COCO | CIDEr: 89.2 -> 91.2 | Helped |
| beam search k=3 instead of greedy decoding (ablation) | GLA · MS COCO | CIDEr: 91.2 -> 96.4 | Helped |
| GLA vs ATT | captioner · MS COCO | BLEU-4: 30.4 -> 31.2 | Helped |
Closest papers by meaning
- Improving Image Captioning by Leveraging Intra- and Inter-layer Global Representation in Transformer Network (AAAI 2021 · similarity 0.87)
- Normalized and Geometry-Aware Self-Attention Network for Image Captioning (CVPR 2020 · similarity 0.86)
- Text-Guided Attention Model for Image Captioning (AAAI 2017 · similarity 0.86)
- Attention Correctness in Neural Image Captioning (AAAI 2017 · similarity 0.85)
- Image Captioning With Semantic Attention (CVPR 2016 · similarity 0.85)
- Patch Matters: Training-free Fine-grained Image Caption Enhancement via Local Perception (CVPR 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.