Summary of Morevqa: Exploring Modular Reasoning Models For Video Question Answering, by Juhong Min et al.
MoReVQA: Exploring Modular Reasoning Models for Video Question Answering
by Juhong Min, Shyamal Buch, Arsha Nagrani, Minsu Cho, Cordelia Schmid
First submitted to arxiv on: 9 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper presents a modular reasoning framework for video question answering (videoQA), which outperforms previous methods on standard benchmarks. The proposed approach, MoReVQA, decomposes the task into three stages: event parsing, grounding, and final reasoning. Each stage is trained using few-shot prompting of large models, producing interpretable intermediate outputs. By breaking down the planning and task complexity, MoReVQA achieves state-of-the-art results on NExT-QA, iVQA, EgoSchema, and ActivityNet-QA benchmarks, as well as extensions to grounded videoQA and paragraph captioning. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Video question answering is a challenge in artificial intelligence that requires a machine learning model to understand videos and answer questions about them. This paper proposes a new approach called MoReVQA, which works by breaking down the task into smaller stages. Each stage uses large language models and is trained using only a few examples of correct answers. The final result is a state-of-the-art video question answering system that can also be used for other related tasks. |
Keywords
* Artificial intelligence * Few shot * Grounding * Machine learning * Parsing * Prompting * Question answering
