ReefScan underwater imagery from nine Great Barrier Reef surveys, with per-image quality assessments (UNIQAT)

Updated: 2026-09-23

Downward-facing underwater imagery collected by AIMS ReefScan systems across nine surveys of the Great Barrier Reef between January 2024 and December 2025, together with per-image quality assessments produced by the UNIQAT toolkit. The collection comprises 130,845 images from three reef sectors (Swains, the outer-shelf Pompey Reefs area, and Townsville), four field trips, and three ReefScan hardware configurations. Each image carries GPS position, sonar depth and timestamp from the instrument photo log, and 37 quantitative image quality metrics computed by the UNIQAT traditional pipeline, including colour cast and blue-water severity, four contrast measures, sharpness and blur, feature richness, visibility and turbidity, the established UCIQE and UIQM underwater metrics, entropy and colour distribution descriptors, and two composite scores for computer vision usefulness and marine science value. The dataset supports reproducible assessment of image quality in marine survey programmes, quality-based triage of large image collections before annotation or analysis, and benchmarking of automated quality assessment methods. Data format and volume: 130,845 images, 623.6 GB total. 126,618 JPEG images at 5312 x 3040 or 5328 x 3040 pixels (eight surveys). 4,227 uncompressed BMP images at 5328 x 3040 pixels (Townsville, Gen2 configuration, 48.6 MB per frame and 205.4 GB for that survey alone). Derived tables: nine per-image CSV files (one per survey) and one per-survey summary CSV, approximately 60 MB uncompressed, also provided as a single archive of a few tens of MB. Keywords: coral reef; Great Barrier Reef; underwater imagery; image quality assessment; ReefScan; marine monitoring; benthic survey; autonomous imaging; data quality