REPID: Rendering Evaluation of Photographic Image Dataset for Aesthetic Preference

REPID is a large-scale benchmark for Image Rendering Quality Assessment (IRQA) designed to model subjective human aesthetic preferences for different rendering styles of the same scene. It was created by author vsevolodpl and was last updated in March 2026.

Use Cases

• Train models to predict aesthetic preference scores for different rendering styles of the same scene.
• Benchmark image rendering algorithms against human subjective preferences for rendering styles.
• Analyze the relationship between specific rendering style parameters and human aesthetic judgments.

Strengths

• Designed as a large-scale benchmark for a specialized task (Image Rendering Quality Assessment).
• Focuses on modeling subjective human aesthetic preferences, unlike traditional technical IQA.

Limitations

• Specific dataset size, file formats, and column structure are unknown from the provided input.
• The dataset's composition, such as the number of scenes or rendering styles, is not detailed.

Provenance

Source

huggingface.co/datasets/vsevolodpl/REPID

Freshness

Last updated on 2026-03-18.