Research map / Vision-language and multimodal / Human and embodied modeling
Image Captioning With Semantic Attention
Check 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 semantic attention over FCN-predicted attributes | LSTM image captioner · Flickr30k | BLEU-4: 0.199 -> 0.230 | Helped |
| add semantic attention over FCN-predicted attributes | LSTM image captioner · MS-COCO | BLEU-4: 0.250 -> 0.304 | Helped |
| add semantic attention over FCN-predicted attributes | LSTM image captioner · Flickr30k | BLEU-1: 0.669 -> 0.647 | Hurt |
| replace ground-truth-attribute ATT with element-wise MAX fusion (ablation) | Ours-GT-ATT · Flickr30k | CIDEr: 0.949 -> 0.747 | Hurt |
| replace ground-truth-attribute ATT with concatenation fusion (ablation) | Ours-GT-ATT · Flickr30k | CIDEr: 0.949 -> 0.685 | Hurt |
| replace ground-truth-attribute ATT with element-wise MAX fusion (ablation) | Ours-GT-ATT · MS-COCO | CIDEr: 1.685 -> 1.161 | Hurt |
| replace ground-truth-attribute ATT with concatenation fusion (ablation) | Ours-GT-ATT · MS-COCO | CIDEr: 1.685 -> 1.237 | Hurt |
| combine input and output attention instead of using input attention alone (ablation) | LSTM captioner with ground-truth attributes · MS-COCO | CIDEr: 1.56 -> 1.68 | Helped |
| combine input and output attention instead of using output attention alone (ablation) | LSTM captioner with ground-truth attributes · MS-COCO | CIDEr: 1.58 -> 1.68 | Helped |
| use FCN rather than k-NN or ranking-loss attribute detection with ATT (ablation) | semantic-attention captioner · MS-COCO | BLEU-4: 0.281 / 0.282 -> 0.304 | Helped |
| replace ATT with CON or MAX for FCN-predicted attributes (ablation) | semantic-attention captioner · MS-COCO | BLEU-4: 0.304 -> 0.300 / 0.301 | Hurt |
Closest papers by meaning
- Text-Guided Attention Model for Image Captioning (AAAI 2017 · similarity 0.90)
- A Hierarchical Approach for Generating Descriptive Image Paragraphs (CVPR 2017 · similarity 0.89)
- Convolutional Image Captioning (CVPR 2018 · similarity 0.88)
- Aligning Visual Regions and Textual Concepts for Semantic-Grounded Image Representations (NeurIPS 2019 · similarity 0.88)
- Top-Down Semantic Refinement for Image Captioning (AAAI 2026 · similarity 0.87)
- Noise-Aware Image Captioning with Progressively Exploring Mismatched Words (AAAI 2024 · similarity 0.87)
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.