Research map / Vision-language and multimodal / Visual question answering
Visual Question Generation as Dual Task of Visual Question Answering
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 |
|---|---|---|---|
| Jointly train VQA and VQG with dual training | iBOWIMG · VQA2 subset | Acc@1: 42.05 -> 43.44 | Helped |
| Jointly train VQA and VQG with dual training | MLB · VQA2 subset | Acc@1: 50.23 -> 50.83 | Helped |
| Jointly train VQA and VQG with dual training | MUTAN + SkipThought · VQA2 subset | Acc@1: 50.72 -> 51.49 | Helped |
| Jointly train VQA and VQG with dual training | MUTAN + LSTM · VQA2 subset | BLEU: 54.17 -> 54.89 | Helped |
| Add Dual Mutan (ablation) | iQAN · VQA2 subset | Acc@1: 50.72 -> 50.99 | Helped |
| Add shared Q/A encoders and decoders (ablation) | iQAN with Dual Mutan · VQA2 subset | Acc@1: 50.99 -> 51.23 | Helped |
| Add duality regularizer (ablation) | iQAN with Dual Mutan · VQA2 subset | Acc@1: 50.99 -> 51.38 | Helped |
| Add duality regularizer (ablation) | iQAN with Dual Mutan · VQA2 Full | Acc@1: 54.78 -> 54.39 | Hurt |
| Add shared Q/A encoders and decoders (ablation) | iQAN with Dual Mutan and duality regularizer · CLEVR | Overall Acc@1: 84.78 -> 85.07 | Helped |
| Use VQG-generated questions for training, then fine-tune on labeled Q/A pairs | VQG + dual training · VQA2 with 0.1 <Q,A> + 0.9 A | Acc@1: 35.23 -> 39.95 | Helped |
| Use VQG-generated questions for training, then fine-tune on labeled Q/A pairs | VQG + dual training · VQA2 with 0.1 <Q,A> + 0.9 A | BLEU: 48.45 -> 49.40 | Helped |
| Train with VQG-generated questions without fine-tuning on labeled Q/A pairs | VQG + dual training · VQA2 with 0.1 <Q,A> + 0.9 A | BLEU: 48.45 -> 44.92 | Hurt |
Closest papers by meaning
- Multi-Question Learning for Visual Question Answering (AAAI 2020 · similarity 0.92)
- Answer Them All! Toward Universal Visual Question Answering Models (CVPR 2019 · similarity 0.91)
- VQ-VA World: Towards High-Quality Visual Question-Visual Answering (CVPR 2026 · similarity 0.91)
- Tips and Tricks for Visual Question Answering: Learnings From the 2017 Challenge (CVPR 2018 · similarity 0.91)
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.90)
- ConVQG: Contrastive Visual Question Generation with Multimodal Guidance (AAAI 2024 · similarity 0.89)
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.