WebUI-COCO-868: Annotated Webpage Screenshots for UI Detection

868 desktop webpage screenshots are annotated with COCO bounding boxes for ARIA landmark regions and common UI components. The dataset, created by jileklu and last updated in January 2026, provides a resource for training models to identify elements like cookie dialogs and buttons in web interfaces.

Use Cases

• Train object detection models to locate UI components like buttons and popovers based on the annotated bounding boxes.
• Develop models for web accessibility by detecting ARIA landmark regions mentioned in the description.
• Benchmark UI element segmentation and detection algorithms using the provided COCO-style annotations.
• Analyze the prevalence and placement of common interface elements like cookie dialogs and captchas across webpages.

Strengths

• Contains 868 annotated images, providing a substantive collection for model training.
• Annotations include COCO bounding boxes for both ARIA landmarks and specific UI components, offering dual utility.
• Last updated on 2026-01-03, suggesting recent maintenance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset's scale is limited to 868 images, which may not capture the full diversity of web interfaces.
• Data may reflect source bias inherent to the specific webpages captured.

Provenance

Source

huggingface user jileklu

Collection Method

Likely involves screenshot capture and manual or semi-automated annotation of desktop webpages.

Time Range

null

Freshness

Last updated 2026-01-03 02:18:46.

Geography

null