VideoInstruct (Video Instruction Dataset)
Introduced by Maaz et al. in Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Models
Video Instruction Dataset is used to train Video-ChatGPT. It consists of 100,000 high-quality video instruction pairs. employs a combination of human-assisted and semi-automatic annotation techniques, aiming to produce high-quality video instruction data. These methods create question-answer pairs related to
- Video summarization
- Description-based question-answers (exploring spatial, temporal, relationships, and reasoning concepts)
- Creative/generative question-answers
The details are available at https://github.com/mbzuai-oryx/Video-ChatGPT/blob/main/data/README.md.
Benchmarks
| Trend | Task | Dataset Variant | Best Model | Paper | Code |
|---|---|---|---|---|---|
|
VideoInstruct
|
LLaMA-VID-13B
|
|||
|
VideoInstruct
|
VideoChat2
|
|||
|
VideoInstruct
|
Chat-UniVi
|
|||
|
VideoInstruct
|
VideoChat2
|
|||
|
VideoInstruct
|
VTimeLLM
|
|||
|
VideoInstruct
|
Chat-UniVi
|
Papers
| Paper | Code | Results | Date | Stars |
|---|
Dataset Loaders
No data loaders found. You can
submit your data loader here.
No data loaders found. You can
submit your data loader here.
