Research map / Video / Video-language understanding
Goldfish: Vision-Language Understanding of Arbitrarily Long Videos
Kirolos Ataallah, Xiaoqian shen, Eslam mohamed abdelrahman, Essam Sleiman, Mingchen Zhuge, Jian Ding, Deyao Zhu, Jürgen Schmidhuber, Mohamed Elhoseiny
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 the retrieval module over clip descriptions instead of feeding 45 sampled frames (ablation) | Goldfish · TVQA-Long | accuracy: 25.07% -> 41.78% | Improved |
| use generated summaries in addition to subtitles for retrieval (ablation) | Goldfish retrieval · TVQA-Long | retrieval accuracy: 39.7 -> 39.5 | No clear effect |
| retrieve with generated summaries rather than subtitles (ablation) | Goldfish retrieval · TVR-Vision | retrieval accuracy: 48.4 -> 51.2 | Improved |
| use OpenAI text-embedding-3-small instead of bert-base-nli-mean-tokens (ablation) | Goldfish · TVQA-Long | overall accuracy: 28.4 -> 41.78 | Improved |
| feed retrieved summaries and subtitles directly to the LLM instead of re-running MiniGPT4-Video on retrieved clips (ablation) | Goldfish answer module · TVQA-Long | accuracy: 27.6 -> 41.78 | Improved |
| increase video length from 6 to 24 minutes by aggregating more clips (ablation) | Goldfish · TVQA (5% val subset) | retrieval and overall accuracy | No clear effect |
| add subtitles to visual input (ablation) | Goldfish · TVQA-Long | accuracy: 28.61 -> 41.78 | Improved |
| add subtitles to visual input (ablation) | MiniGPT4-Video · TVQA (short) | accuracy: 36.45 -> 46.94 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- LLaMA-VID — LLaMA-VID: An Image is Worth 2 Tokens in Large Language Models (ECCV 2024) · better on 7 of 8 shared results
- MovieChat — MovieChat: From Dense Token to Sparse Memory for Long Video Understanding (CVPR 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Streaming Long Video Understanding with Large Language Models (NeurIPS 2024 · similarity 0.91)
- Video-RAG: Visually-aligned Retrieval-Augmented Long Video Comprehension (NeurIPS 2025 · similarity 0.90)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.89)
- Vgent: Graph-based Retrieval-Reasoning-Augmented Generation For Long Video Understanding (NeurIPS 2025 · similarity 0.89)
- Unleashing Hour-Scale Video Training for Long Video-Language Understanding (NeurIPS 2025 · similarity 0.89)
- MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video 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.
