Loading Now

Summary of Traveler: a Modular Multi-lmm Agent Framework For Video Question-answering, by Chuyi Shang et al.


TraveLER: A Modular Multi-LMM Agent Framework for Video Question-Answering

by Chuyi Shang, Amos You, Sanjay Subramanian, Trevor Darrell, Roei Herzig

First submitted to arxiv on: 1 Apr 2024

Categories

  • Main: Computer Vision and Pattern Recognition (cs.CV)
  • Secondary: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)

     Abstract of paper      PDF of paper


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 proposes a modular multi-Large Multimodal Models (LMM) agent framework to improve video question-answering (VideoQA). The current frame-wise approach can’t adapt if insufficient or incorrect information is collected. To overcome this, the authors introduce TraveLER, a method that creates a plan, asks questions about individual frames to locate and store key information, evaluates if there’s enough information to answer the question, and replans based on its collected knowledge. The proposed approach improves performance on several VideoQA benchmarks without fine-tuning on specific datasets.
Low GrooveSquid.com (original content) Low Difficulty Summary
The paper tries to make video-based question-answering better. Right now, it’s like having one superpower that can do everything – find information, understand what you need, and give the answer all at once. But this makes mistakes more likely. The authors suggest a team of smaller “superpowers” working together to get better results. Each one has its own job: finding important parts in the video, asking questions about those parts, checking if they have enough information, and redoing their plan if needed. This helps them answer questions correctly without needing special training for each task.

Keywords

» Artificial intelligence  » Fine tuning  » Question answering