Research map / Label-efficient and robust learning / Domain adaptation and generalization
Show, Adapt and Tell: Adversarial Training of Cross-Domain Image Captioner
Tseng-Hung Chen, Yuan-Hong Liao, Ching-Yao Chuang, Wan-Ting Hsu, Jianlong Fu, Min Sun
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 multi-modal critic only (ablation) | Source Pre-trained captioner · MSCOCO to Flickr30k (ablation) | CIDEr-D: 25.3 -> 25.9 | No clear effect |
| add domain critic only (ablation) | Source Pre-trained captioner · MSCOCO to Flickr30k (ablation) | CIDEr-D: 25.3 -> 32.1 | Improved |
| add multi-modal critic to domain critic (ablation) | +DC · MSCOCO to Flickr30k (ablation) | CIDEr-D: 32.1 -> 32.6 | Improved |
| adversarial adaptation with unpaired target data | Source Pre-trained captioner · MSCOCO to CUB-200 | CIDEr: 3 -> 24.8 | Improved |
| adversarial adaptation with unpaired target data | Source Pre-trained captioner · MSCOCO to Oxford-102 | CIDEr: 3.1 -> 29.3 | Improved |
| adversarial adaptation with unpaired target data | Source Pre-trained captioner · MSCOCO to TGIF | CIDEr: 14.7 -> 22.2 | Improved |
| adversarial adaptation with unpaired target data | Source Pre-trained captioner · MSCOCO to Flickr30k | CIDEr: 25.3 -> 32.6 | Improved |
| reverse adaptation direction, Flickr30k to MSCOCO | Source Pre-trained captioner · Flickr30k to MSCOCO | CIDEr: 27.3 -> 38.2 | Improved |
| critic-based planning instead of greedy search | Ours · MSCOCO to CUB-200 | CIDEr-D: 24.8 -> 29.3 | Improved |
| critic-based planning instead of greedy search | Ours · MSCOCO to CUB-200 | BLEU-4: 32.8 -> 35.2 | Improved |
| critic-based planning instead of greedy search | Ours · MSCOCO to Oxford-102 | CIDEr-D: 29.3 -> 24.9 | Hurt |
| critic-based planning instead of greedy search | Ours · MSCOCO to Oxford-102 | BLEU-4: 60.5 -> 62.4 | Improved |
Closest papers by meaning
- Cross-Domain Image Captioning With Discriminative Finetuning (CVPR 2023 · similarity 0.89)
- Adversarial Semantic Alignment for Improved Image Captions (CVPR 2019 · similarity 0.89)
- DECIDER: Difference-aware Contrastive Diffusion Model with Adversarial Perturbations for Image Change Captioning (AAAI 2025 · similarity 0.88)
- Speaking the Same Language: Matching Machine to Human Captions by Adversarial Training (ICCV 2017 · similarity 0.88)
- Improving Image Captioning with Conditional Generative Adversarial Nets (AAAI 2019 · similarity 0.88)
- CapRL: Stimulating Dense Image Caption Capabilities via Reinforcement Learning (ICLR 2026 · 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.
