Research map / Vision-language and multimodal / Human and embodied modeling
Show, Control and Tell: A Framework for Generating Controllable and Grounded Captions
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 explicit region-sequence control and noun-chunk grounding | Show, Control and Tell · COCO Entities test, sequence control | NW: 0.443 -> 0.649 | Helped |
| add explicit region-sequence control and noun-chunk grounding | Show, Control and Tell · Flickr30k Entities test, sequence control | CIDEr: 69.5 -> 84.0 | Helped |
| use attention and a stronger representation of the control signal instead of a single visual feature | Controllable LSTM · COCO Entities test, sequence control, CIDEr + NW optimization | CIDEr: 43.5 -> 69.5 | Helped |
| optimize with CIDEr and NW alignment rewards rather than CIDEr alone | Show, Control and Tell · COCO Entities test, sequence control | NW: 0.615 -> 0.649 | Helped |
| add the visual sentinel rather than merging it with the chunk sentinel (ablation) | Show, Control and Tell · sequence-controlled COCO Entities and Flickr30k Entities | CIDEr: 201.1 -> 209.7; 76.5 -> 84.0 | Helped |
| add the visual sentinel rather than removing it (ablation) | Show, Control and Tell · sequence-controlled Flickr30k Entities | CIDEr: 74.7 -> 84.0 | Helped |
| add the visual sentinel rather than removing it (ablation) | Show, Control and Tell · sequence-controlled COCO Entities | CIDEr: 202.2 -> 209.7 | Helped |
| add the visual sentinel rather than removing it (ablation) | Show, Control and Tell · set-controlled COCO Entities and Flickr30k Entities | CIDEr: 168.9 -> 173.3; 64.2 -> 70.4 | Helped |
| add the visual sentinel rather than merging it with the chunk sentinel (ablation) | Show, Control and Tell · set-controlled COCO Entities and Flickr30k Entities | CIDEr: 168.4 -> 173.3; 65.6 -> 70.4 | Helped |
| use the learned Sinkhorn ranking instead of a predefined global class ranking | Region sorting network · COCO Entities | Accuracy: 59.1% -> 67.1% | Helped |
| use the learned Sinkhorn ranking instead of a predefined global class ranking | Region sorting network · Flickr Entities | Accuracy: 58.4% -> 65.2% | Helped |
Closest papers by meaning
- Human-Like Controllable Image Captioning With Verb-Specific Semantic Roles (CVPR 2021 · similarity 0.89)
- Controllable Image Captioning via Prompting (AAAI 2023 · similarity 0.89)
- Cycle-Consistency Learning for Captioning and Grounding (AAAI 2024 · similarity 0.87)
- Panoptic Captioning: An Equivalence Bridge for Image and Text (NeurIPS 2025 · similarity 0.86)
- FlexCap: Describe Anything in Images in Controllable Detail (NeurIPS 2024 · similarity 0.86)
- Set Prediction Guided by Semantic Concepts for Diverse Video Captioning (AAAI 2024 · similarity 0.86)
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.