NESP TWQ 5.5: 3,500 Great Barrier Reef Image Aesthetic Ratings for AI Training

The NESP TWQ 5.5 project from the Griffith Institute for Tourism Research created a dataset to improve an AI system for assessing the beauty of natural scenes. It contains 3,120 underwater images of the Great Barrier Reef sourced from Flickr and private collections, rated for aesthetic beauty on a 10-point scale by 1,585 Australian survey participants in April 2020. The dataset includes a report, word files, and an Excel file with the collected ratings.

Use Cases

• Training AI models for automatic aesthetic scoring of natural scenes based on human-rated image data.
• Analyzing public perception of marine landscape beauty based on survey ratings collected in 2020.
• Improving the accuracy of automated aesthetic monitoring systems for environmental research.

Strengths

• Includes 3,120 specifically selected underwater images of the Great Barrier Reef.
• Contains aesthetic ratings from 1,585 complete and valid survey responses.
• Builds upon prior research from the NESP TWQ Hub Project 3.2.3, suggesting a research lineage.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the rating data is unknown, which may limit suitability assessment.
• Data may reflect geographic and temporal bias, being focused on the Great Barrier Reef and collected in 2020.

Provenance

Source

Griffith Institute for Tourism Research, NESP TWQ 5.5 project, via Australian Ocean Data Network.

Collection Method

Images were downloaded from Flickr using a Python application with the Flickr API and selected from a pool of approximately 55,000 images. Aesthetic ratings were collected via an online Qualtrics survey.

Time Range

Survey launched in April 2020.

Freshness

Last updated 2026-07-16 01:42:15.623916; freshness should be verified.

Geography

Great Barrier Reef, Australia.