AI Tool Revolutionizes Forensic Sex Estimation with 97% Accuracy

Developed by CSIRO and University of Western Australia, the AI model outperforms human assessors and traditional methods in speed and precision.

The AI tool, created by CSIRO and University of Western Australia, analyzes skulls to estimate biological sex with 97% accuracy, surpassing the 82% accuracy of human assessors.
It processes results five times faster than humans, providing quicker answers in forensic investigations, such as in mass casualty events or natural disasters.
The model was trained on a dataset of 200 CT scans from Indonesia, addressing limitations in traditional methods that often rely on population-specific skull traits.
Unlike the conventional Walker method, the AI identifies sex-associated traits by learning new features, focusing on overall skull morphology rather than just five predefined traits.
Researchers are seeking industry collaborators to expand the dataset for greater diversity and to adapt the tool for real-world forensic applications.