ColorSenseNet: A Benchmark for Color Similarity Learning on Fabric Textures

58,000+ high-resolution images of fabric textures with color variations, created by Muzammil Hasan and last updated in April 2026. The dataset includes 30 base fabric images, each processed to generate around 2000 color variations via HSV adjustments, covering over 590 unique colors. 60% of the data is single contrast plain fabric and 40% is double contrast plain fabric.

Use Cases

• Training color similarity models based on the 590+ unique color variations.
• Benchmarking texture-preserving image augmentation techniques based on HSV adjustments.
• Developing models for realistic fabric shade generation based on the described processing pipeline.

Strengths

• 58,000+ images provide a substantial volume for model training.
• Over 590 unique colors derived from 30 base fabric types offer diversity.
• High-resolution base images with preserved texture detail are mentioned.
• Explicit breakdown of 60% single contrast and 40% double contrast fabric types.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect bias inherent to the specific fabric types and color adjustment methods used.

Provenance

Source

Muzammil Hasan via figshare

Collection Method

30 high-resolution base images of fabric were processed to generate color variations using HSV adjustments.

Freshness

Last updated 2026-04-20 11:59:32; freshness should be verified.