Research map / Large language models / Agents, code and math
DoraemonGPT: Toward Understanding Dynamic Scenes with Large Language Models (Exemplified as A Video Agent)
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes VVGPT (DoraemonGPT), an LLM-driven video agent that converts videos into a symbolic memory, uses multi-source knowledge with a router, and runs tree-of-thought planning to answer complex dynamic-scene questions.
Problem
LLM-driven visual agents focus on image tasks and lack the ability to understand dynamic video scenes needed for real-world applications such as guiding laboratory experiments.
Approach
Converts input video into a flexible structured symbolic memory for retrieval, combines multi-source domain knowledge with a router, and automates execution via tree-like planning that explores multiple reasoning paths.
Key results
- The paper states VVGPT is validated on a synthetic dataset for complex reasoning in real-world scenarios, but the provided text gives no numeric results.
Limitations
- The provided text is an incomplete draft with placeholder sections (e.g., 'haha.'), so method and results details are missing.
Experimental setup
Benchmarks: synthetic dataset for complex real-world reasoning
Builds on: tree of thoughts ChatGPT
Closest papers by meaning
- VideoRefer Suite: Advancing Spatial-Temporal Object Understanding with Video LLM (CVPR 2025 · similarity 0.87)
- AdaReasoner: Dynamic Tool Orchestration for Iterative Visual Reasoning (ICLR 2026 · similarity 0.87)
- OST-Bench: Evaluating the Capabilities of MLLMs in Online Spatio-temporal Scene Understanding (NeurIPS 2025 · similarity 0.87)
- Video-of-Thought: Step-by-Step Video Reasoning from Perception to Cognition (ICML 2024 · similarity 0.87)
- MATA: A Trainable Hierarchical Automaton System for Multi-Agent Visual Reasoning (ICLR 2026 · similarity 0.87)
- NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Models (AAAI 2024 · similarity 0.86)
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.