VideoInstruct100K Dataset Overview

Dataset Description

VideoInstruct100K is a high‑quality video‑dialogue dataset generated using human‑in‑the‑loop and semi‑automatic annotation techniques. The Q&A content includes the following aspects:

• Video summarization
• Description‑based question answering (exploring spatial, temporal, relational, and reasoning concepts)
• Creative/generative question answering

Citation Information

If you find this dataset useful, please consider citing the following paper:

@article{Maaz2023VideoChatGPT,
    title={Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Models},
    author={Muhammad Maaz, Hanoona Rasheed, Salman Khan and Fahad Khan},
    journal={ArXiv 2306.05424},
    year={2023}
}