Danbooru2024: Annotated Anime and Illustration Images

Deepghs created the Danbooru2024 Webp 4MPixel dataset, derived from the Danbooru platform. It contains approximately 8.05 million high-quality images of animation and illustration artwork. The dataset was last updated on November 18, 2024.

Use Cases

• Training image classification models based on user-annotated tags.
• Developing image generation models for anime and illustration styles.
• Researching multimodal AI that links visual content with textual descriptions.
• Analyzing trends in fan-created artwork based on the large collection.

Strengths

• Approximately 8.05 million images provide a large-scale resource.
• Images are described as high-quality and user-annotated with tags and descriptions.
• The dataset is focused on a specific domain of animation and illustration artwork.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Official Danbooru platform.

Collection Method

Derived and resized from the original Danbooru2024 dataset.

Freshness

Last updated 2024-11-18 13:26:12; freshness should be verified.