Research map / Video / Video-language understanding
Video Question Answering Using Language-Guided Deep Compressed-Domain Video Feature
Nayoung Kim, Seong Jong Ha, Je-Won Kang
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 compressed-domain motion vectors (ablation) | VQAC(Base) · MSVD-QA | top-1 accuracy: 1.1% drop without MV | Improved |
| add compressed-domain motion vectors (ablation) | VQAC(HME) · MSVD-QA | top-1 accuracy: 3.5% drop without MV | Improved |
| add residual features (ablation) | VQAC(Base) · MSVD-QA | top-1 accuracy: 0.4% drop without residual | Improved |
| add residual features (ablation) | VQAC(HME) · MSVD-QA | top-1 accuracy: 1.2% drop without residual | Improved |
| use proposed motion feature instead of compressed-domain action-recognition feature (ablation) | VQAC(Base) · MSVD-QA | top-1 accuracy: 0.7% performance drop with DMC | Improved |
| use proposed motion feature instead of SlowFast (ablation) | VQAC(Base) · MSVD-QA | top-1 accuracy: 0.2% drop with SF | Improved |
| add global question feature sw (ablation) | VQAC(HME) · MSVD-QA | top-1 accuracy: 2.3% drop without sw | Improved |
| use motion vectors instead of 3D-CNN motion feature (ablation) | VQAC(HME) · MSVD-QA | top-1 accuracy: 1.7% drop with 3D | Improved |
| add HME memory to VQAC | VQAC(Base) · MSVD-QA | top-1 accuracy: 31.5 -> 37.8 | Improved |
| replace 3D-CNN motion features with compressed-domain features | HME · MSVD-QA | total inference time (min): 11.9 -> 5.5 | Improved |
Closest papers by meaning
- Structured Two-Stream Attention Network for Video Question Answering (AAAI 2019 · similarity 0.90)
- Leveraging Video Descriptions to Learn Video Question Answering (AAAI 2017 · similarity 0.89)
- BIMBA: Selective-Scan Compression for Long-Range Video Question Answering (CVPR 2025 · similarity 0.88)
- Language-aware Visual Semantic Distillation for Video Question Answering (CVPR 2024 · similarity 0.88)
- Streaming Long Video Understanding with Large Language Models (NeurIPS 2024 · similarity 0.88)
- VideoEspresso: A Large-Scale Chain-of-Thought Dataset for Fine-Grained Video Reasoning via Core Frame Selection (CVPR 2025 · similarity 0.88)
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.
