Spicule Measurement For Marine Sponge Taxonomy

Updated: 2026-05-20

This dataset supports the development and validation of an AI-assisted workflow for marine sponge spicule measurement and taxonomic analysis. Sponge taxonomy relies heavily on manual spicule morphometrics, which is time consuming and difficult to scale, so the data were collected to create expert validated reference material for training and benchmarking computer vision models. The dataset includes high-resolution microscope imagery, expert annotations, segmentation outputs, and morphometric measurements of sponge spicules, including oxea and tylostyle forms. Images were generated from slide mounted sponge spicule preparations using brightfield and relief microscopy at multiple magnifications, with data collected and processed during 2025-2026 through collaboration between AIMS and University of Western Australia taxonomic specialists. The data are being interpreted using AI-based image segmentation and measurement workflows to extract spicule length, width, and statistical descriptors, which are compared against expert derived measurements to assess model performance and support taxonomic interpretation.