ServImage: Commercial Design Tasks with Human Acceptance Labels

ServImage is a benchmark dataset for image generation and editing tasks in real-world commercial design services. According to the paper, it contains about 1,070 design tasks and 2,046 deliveries, expanded to roughly 33,000 candidate images. The dataset was created by FengxianUP and was last updated on Hugging Face in January 2026.

Use Cases

• Benchmarking image generation models based on real-world commercial design tasks.
• Training models to predict human acceptance (pay-worthiness) of generated images.
• Evaluating image editing techniques for product, portrait, and digital design categories.

Strengths

• Contains about 1,070 commercial design tasks and 2,046 deliveries, providing a substantial task base.
• Includes roughly 33,000 candidate images with associated human feedback signals.
• Covers three distinct commercial design categories: Portrait, Product, and Digital.

Limitations

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

Provenance

Source

huggingface

Freshness

Last updated 2026-01-31 13:52:41; freshness should be verified.