Research map / Vision-language and multimodal / Visual question answering
TGIF-QA: Toward Spatio-Temporal Reasoning in Visual Question Answering
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 |
|---|---|---|---|
| Use both ResNet and C3D visual features instead of text only (ablation) | ST-VQA · TGIF-QA repeating action | accuracy: 47.91 -> 60.13 | Helped |
| Use both ResNet and C3D visual features instead of text only (ablation) | ST-VQA · TGIF-QA repetition count | mean ℓ2 loss: 5.0056 -> 4.3759 | Helped |
| Add ResNet features to C3D features (ablation) | ST-VQA · TGIF-QA repetition count | mean ℓ2 loss: 4.4478 -> 4.3759 | Helped |
| Add spatial attention without temporal attention (ablation) | ST-VQA-Concat · TGIF-QA repetition count | mean ℓ2 loss: 4.3759 -> 4.2825 | Helped |
| Add spatial attention without temporal attention (ablation) | ST-VQA-Concat · TGIF-QA repeating action | accuracy: 60.13 -> 57.33 | Hurt |
| Add temporal attention without spatial attention (ablation) | ST-VQA-Concat · TGIF-QA state transition | accuracy: 65.70 -> 67.06 | Helped |
| Add temporal attention without spatial attention (ablation) | ST-VQA-Concat · TGIF-QA repetition count | mean ℓ2 loss: 4.3759 -> 4.3981 | Hurt |
| Combine spatial and temporal attention (ablation) | ST-VQA-Concat · TGIF-QA state transition | accuracy: 65.70 -> 59.59 | Hurt |
| Use temporal attention rather than the strongest reported non-proposed baseline | TGIF-QA repeating action · TGIF-QA repeating action | accuracy: 58.85 -> 60.77 | Helped |
Closest papers by meaning
- When Open-Vocabulary Visual Question Answering Meets Causal Adapter: Benchmark and Approach (AAAI 2025 · similarity 0.91)
- Track the Answer: Extending TextVQA from Image to Video with Spatio-Temporal Clues (AAAI 2025 · similarity 0.91)
- SpatiaLQA: A Benchmark for Evaluating Spatial Logical Reasoning in Vision-Language Models (CVPR 2026 · similarity 0.90)
- Not Another Text Benchmark: Putting the “Visual" Back in Visual Question Answering for Large Video Models (NeurIPS 2026 · similarity 0.90)
- Language-aware Visual Semantic Distillation for Video Question Answering (CVPR 2024 · similarity 0.90)
- Tool-Augmented Spatiotemporal Reasoning for Streamlining Video Question Answering Task (NeurIPS 2025 · similarity 0.90)
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.