Website Aesthetics: 3,800+ Rated Web Screenshots for ML

Created by calista-ai and last updated in March 2023, this repository provides datasets focused on website aesthetics assessment using crowdsourced pairwise comparisons and rating-based ranking. The data supports the development of machine learning models for image aesthetics assessment specifically within the domain of web design.

Use Cases

• Training Bradley-Terry models to rank website designs based on pairwise comparison labels
• Developing image aesthetics assessment (IAA) algorithms for automated web design evaluation
• Benchmarking rating-based ranking systems using crowdsourced aesthetic scores

Strengths

• Includes pairwise comparison data suitable for Bradley-Terry modeling
• Focuses specifically on website aesthetics rather than general photography
• Released under a permissive MIT license for open research and commercial use

Limitations

• Unknown record count and file size
• Subjective bias inherent in crowdsourced aesthetic ratings
• Lack of explicit column definitions in the primary metadata

Provenance

Source

calista-ai GitHub repository

Collection Method

Crowdsourcing

Freshness

Last updated March 28, 2023.