Research map / Vision-language and multimodal / Human and embodied modeling
Hierarchical Attention Network for Image Captioning
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 |
|---|---|---|---|
| use hierarchical (patch+object+text) features instead of patch features only (ablation) | captioning model without MRM · MSCOCO Karpathy test | CIDEr: 106.8 -> 112.4 | Helped |
| use hierarchical features instead of object features only (ablation) | captioning model without MRM · MSCOCO Karpathy test | CIDEr: 108.3 -> 112.4 | Helped |
| use hierarchical features instead of text features only (ablation) | captioning model without MRM · MSCOCO Karpathy test | CIDEr: 94.8 -> 112.4 | Helped |
| replace addition fusion with multivariate residual module (ablation) | HAN · MSCOCO Karpathy test | CIDEr: 111.2 -> 114.2 | Helped |
| replace concatenation fusion with MRM (ablation) | HAN · MSCOCO Karpathy test | CIDEr: 112.4 -> 114.2 | Helped |
| add parallel MRM with context gate (pMRM) (ablation) | HAN · MSCOCO Karpathy test | CIDEr: 114.2 -> 114.8 | Helped |
| HAN vs NBT under XE training | image captioning · MSCOCO Karpathy test | CIDEr: 107.2 -> 114.8 | Helped |
Closest papers by meaning
- Text-Guided Attention Model for Image Captioning (AAAI 2017 · similarity 0.89)
- Deliberate Attention Networks for Image Captioning (AAAI 2019 · similarity 0.88)
- Normalized and Geometry-Aware Self-Attention Network for Image Captioning (CVPR 2020 · similarity 0.88)
- Attention Correctness in Neural Image Captioning (AAAI 2017 · similarity 0.88)
- Image Captioning With Semantic Attention (CVPR 2016 · similarity 0.88)
- Attention-Aligned Transformer for Image Captioning (AAAI 2022 · 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.