EG-VQA

This repository provides the videos and annotations of EG-VQA. EG-VQA is an evidence-grounded, open-ended Video Question Answering benchmark. It contains 2,067 videos and 11,838 question-answer pairs. Each question is paired with one or more temporally localized evidence segments, including timestamps and textual descriptions.

Questions cover four types:

• Descriptive: recognizing and describing visible events or states.
• Temporal: reasoning about event order, timing, and temporal relations.
• Causal: identifying causes, effects, and event dependencies.
• Counterfactual: reasoning about hypothetical changes to observed events.

Data Sources

EG-VQA is constructed from videos originating from three existing video-language datasets.

Source Dataset	Train Videos	Test Videos	Total Videos
ActivityNet Captions	1,277	223	1,500
YouCook2	168	41	209
HiREST	286	72	358
Total	1,731	336	2,067

The train/test split is performed at the video level, so the same video never appears in both splits.

Dataset Structure

EG-VQA/
├── videos/
│   ├── <video_id>.mp4
│   ├── <video_id>.mkv
│   └── ...
├── train.json
├── test.json
└── README.md

Both train.json and test.json are JSON lists. Each top-level record corresponds to one video and contains the video metadata, source temporal segments, and all EG-VQA questions associated with that video.

Annotation Format

A typical record has the following structure:

{
  "video_id": "example_video_id",
  "video_path": "videos/example_video_id.mp4",
  "duration": 120.0,
  "data_source": "YouCook2",
  "metadata": {
    "title": "Example instructional video",
    "segments": [
      {
        "timestamp": [12.0, 24.5],
        "description": "The person prepares an object on the work surface."
      }
    ]
  },
  "questions": [
    {
      "question_id": "example_video_id_q01",
      "question": "What does the person do after preparing the object?",
      "type": "temporal",
      "answer": "The person places it on the work surface.",
      "evidence": [
        {
          "timestamp": [24.5, 31.0],
          "description": "The prepared object is placed on the work surface."
        }
      ]
    }
  ]
}

Field Description

Video-level fields

Field	Description
video_id	Unique identifier of the video.
video_path	Relative path to the video file under videos/.
duration	Video duration in seconds.
data_source	Original source dataset: ActivityNet Captions, YouCook2, or HiREST.
metadata	Metadata inherited from the source dataset.
metadata.title	Video title when available.
metadata.segments	Temporally localized source descriptions associated with the video.
questions	List of EG-VQA question-answer annotations for the video.

Question-level fields

Field	Description
question_id	Unique question identifier.
question	Open-ended natural-language question.
type	Question type: descriptive, temporal, causal, or counterfactual.
answer	Reference answer.
evidence	One or more evidence segments supporting the answer.
evidence[].timestamp	Temporal interval [start, end] in seconds.
evidence[].description	Natural-language description of the supporting video evidence.

License

The EG-VQA annotations are released under CC BY 4.0. The underlying source videos and source-dataset content remain subject to their respective original licenses and terms of use.

Citation

If you find EG-VQA useful for your research, please cite our paper:

@article{huang2026egvqa,
  title={EG-VQA: Benchmarking Verifiable Video Question Answering with Grounded Temporal Evidence},
  author={Huang, Linpeng and Chen, Weixing and Chen, Zexin and Liu, Yang and Lin, Liang},
  journal={arXiv preprint arXiv:2606.24797},
  year={2026}
}

Acknowledgements

EG-VQA builds on videos and annotations from ActivityNet Captions, YouCook2, and HiREST. We thank the authors and contributors of these datasets for making their resources available to the research community.