Dataset · ServiceNow
Annotated computer-use video recordings
ServiceNow/VideoCUA
55 hours of expert desktop app videos with manual annotations for training computer agents.
What's inside
Contains about 10,000 computer control tasks in English, representing 55 hours of continuous screen recording at 30 fps (6 million frames). Covers 87 professional desktop applications with manual human annotations. Released by the ServiceNow team as part of the CUA-Suite project.
What you can do with it
- Training multimodal agents (VLM / Computer-Use Agents) to control graphical user interfaces
- Training models to understand user action sequences from video streams
- Evaluating accuracy in recognizing UI elements and desktop actions
- Tasks
- Video text → text
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 4.3K
- Likes
- 33
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("ServiceNow/VideoCUA")