Sankaku WebP 4M: Curated Anime-Style Image Subset

Sankaku WebP 4M is a repackaged and curated subset of images selected for training purposes. The dataset, created by MujicaLab, contains anime-style artwork and was last updated on 2025-04-03. Images are organized by ID in tar archives and include filtered tag and image ID lists.

Use Cases

• Train image generation models based on the curated anime-style artwork.
• Develop image classification models based on the filtered tag list.
• Benchmark computer vision algorithms on a subset of images organized by ID.

Strengths

• Curated subset selected for training purposes, suggesting a focus on quality.
• Repackaged version includes filtered tag list (tag_id_mapping.json) and selected image IDs (unique_ids.json).
• Images are organized by ID (0-9) in tar archives, which may aid in data management.

Limitations

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

Provenance

Source

MujicaLab on Hugging Face.

Collection Method

Repackaged version of deepghs/sankaku-webp-4Mpixel.

Freshness

Last updated 2025-04-03 18:29:11; freshness should be verified.