Research map / Video / Video-language understanding
EgoSchema: A Diagnostic Benchmark for Very Long-form Video Language Understanding
Karttikeya Mangalam, Raiymbek Akshulakov, Jitendra Malik
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 |
|---|---|---|---|
| Test humans in the Video → Text setting instead of No constraint | Human QA on EgoSchema · EgoSchema | QA accuracy: 75.0% -> 76.2% | Improved |
| Allow three minutes instead of under one minute | Human QA on EgoSchema · EgoSchema | QA accuracy: +1.0% | Improved |
| Use 1 fps video (180 frames) for human answering | Human QA on EgoSchema · EgoSchema | QA accuracy: 67.2% | Mixed |
| Increase the number of input frames | InternVideo · EgoSchema | QA accuracy | Improved |
| Increase the number of input frames | mPLUG-Owl · EgoSchema | QA accuracy | Mixed |
| Generate questions, correct answers, and wrong answers jointly in two chained calls rather than generating answers in separate stages | Q(AW)-shot generation · EgoSchema question-answer generation | Generated answer quality | Improved |
| Use Q(AW)-shot instead of QAW-shot | Question-answer generation pipeline · EgoSchema question-answer generation | Generation cost: ~30% cheaper | Improved |
| Generate all question-answer triplets in one inference call instead of using staged prompting | One-shot generation · EgoSchema question-answer generation | Generation quality | Hurt |
| Use an LLM to guess the answer from narrations without the question | No-Q baseline · EgoSchema question-answer generation | Answer-guessing accuracy: ~20% | No clear effect |
Closest papers by meaning
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.92)
- HourVideo: 1-Hour Video-Language Understanding (NeurIPS 2024 · similarity 0.91)
- Temporal Chain of Thought: Long-Video Understanding by Thinking in Frames (NeurIPS 2025 · similarity 0.91)
- Omnia de EgoTempo: Benchmarking Temporal Understanding of Multi-Modal LLMs in Egocentric Videos (CVPR 2025 · similarity 0.91)
- CG-Bench: Clue-grounded Question Answering Benchmark for Long Video Understanding (ICLR 2025 · similarity 0.91)
- MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video Understanding (NeurIPS 2024 · similarity 0.90)
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.
