Dataset-interfaces-GUI
Description
This dataset contains 36 screenshots of web user interfaces (UI). The material was selected, provided, and rated by a panel of three subject-matter experts through a cross-evaluation process. This procedure was implemented to establish an inter-rater consensus and group the interfaces into categories based on their visual aesthetics level. The dataset was structured to serve as a testing baseline for research in software engineering and human-computer interaction. Specifically, it enables the comparison between human aesthetic evaluation (obtained via the VisAWI questionnaire) and automated metrics from deep learning models, as well as the analysis of the technology acceptance model (TAM).
Files
Steps to reproduce
The dataset construction and classification process was carried out in three sequential stages: Initial UI Catalog Selection: A panel of three experts with experience in software engineering and UI/UX design provided and selected a preliminary repository of web interface screenshots. Inclusion criteria required the images to be high-resolution captures of main homepages, displaying the full layout of their visual components. Inter-Rater Cross-Evaluation: Each expert independently evaluated the preliminary catalog of interfaces. The analysis focused on core aesthetic attributes such as composition, color harmony, typography, and visual balance. A cross-evaluation matrix was used so that every interface received an individual rating from each of the three judges. Filtering and Consensus on the Final Sample: The individual evaluation scores were cross-referenced. Interfaces were only selected if all three experts reached an unanimous consensus regarding their visual aesthetics tier (high, medium, or low). Any interface that produced conflicting ratings or lacked consensus was excluded. This filtering procedure established the final sample of 36 highly representative web user interfaces that comprise this dataset.
Institutions
- Universidad Nacional Agraria de la SelvaHuánuco Department, Tingo María