Research map / Vision-language and multimodal / Visual question answering
Heterogeneous Memory Enhanced Multimodal Attention Model for Video Question Answering
Chenyou Fan, Xiaofan Zhang, Shu Zhang, Wensheng Wang, Chi Zhang, Heng Huang
Check 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 the heterogeneous visual memory and question memory instead of early feature concatenation (ablation) | HME-VideoQA · MSVD-QA | accuracy: 0.313 -> 0.337 | Improved |
| Add the heterogeneous visual memory and question memory instead of late feature concatenation (ablation) | HME-VideoQA · MSVD-QA | accuracy: 0.315 -> 0.337 | Improved |
| Replace episodic memory with the designed visual memory and add question memory (ablation) | HME-VideoQA · MSVD-QA | accuracy: 0.318 -> 0.337 | Improved |
| Add question memory to visual memory (ablation) | HME-VideoQA · MSVD-QA | accuracy: 0.320 -> 0.337 | Improved |
| Add visual memory to question memory (ablation) | HME-VideoQA · MSVD-QA | accuracy: 0.315 -> 0.337 | Improved |
| Add the heterogeneous visual memory and question memory instead of early feature concatenation (ablation) | HME-VideoQA · MSRVTT-QA | accuracy: 0.309 -> 0.330 | Improved |
| Add the heterogeneous visual memory and question memory instead of late feature concatenation (ablation) | HME-VideoQA · MSRVTT-QA | accuracy: 0.312 -> 0.330 | Improved |
| Replace episodic memory with the designed visual memory and add question memory (ablation) | HME-VideoQA · MSRVTT-QA | accuracy: 0.319 -> 0.330 | Improved |
| Add question memory to visual memory (ablation) | HME-VideoQA · MSRVTT-QA | accuracy: 0.325 -> 0.330 | Improved |
| Add visual memory to question memory (ablation) | HME-VideoQA · MSRVTT-QA | accuracy: 0.321 -> 0.330 | Improved |
| Increase reasoning iterations from 1 to 3 (ablation) | HME-VideoQA · MSVD-QA validation | accuracy: 0.298 -> 0.306 | Improved |
| Increase reasoning iterations from 3 to 5 (ablation) | HME-VideoQA · MSVD-QA validation | accuracy: 0.306 -> 0.307 | Improved |
Closest papers by meaning
- Multimodal Dual Attention Memory for Video Story Question Answering (ECCV 2018 · similarity 0.92)
- MIST: Multi-Modal Iterative Spatial-Temporal Transformer for Long-Form Video Question Answering (CVPR 2023 · similarity 0.91)
- MoReVQA: Exploring Modular Reasoning Models for Video Question Answering (CVPR 2024 · similarity 0.90)
- Modularized Self-Reflected Video Reasoner for Multimodal LLM with Application to Video Question Answering (ICML 2025 · similarity 0.90)
- Language-aware Visual Semantic Distillation for Video Question Answering (CVPR 2024 · similarity 0.89)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 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.
