Stylized Meta Album TEX DTD: Stylized Texture Images for Classification

A stylized version of the Describable Textures Dataset (DTD) is included in the Meta Album collection. The dataset, contributed by author Romain Mussard, is hosted on OpenML and is tagged for machine learning and image benchmarking tasks. Its specific scale and update date are not provided in the available metadata.

Use Cases

• Benchmarking texture classification algorithms based on stylized image data.
• Training models for material recognition under artistic or synthetic transformations.
• Studying model robustness to non-photorealistic, stylized visual inputs.
• Comparing performance between original and stylized versions of texture datasets.

Strengths

• Part of the structured Meta Album collection for machine learning benchmarking.
• Based on the established Describable Textures Dataset (DTD), suggesting a known foundation.
• Explicitly tagged for texture classification and image benchmark tasks on its platform.

Limitations

• Row count, file size, and column details are unknown, limiting suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

OpenML, derived from the Describable Textures Dataset (DTD).

Collection Method

null

Time Range

null

Freshness

null

Geography

null