Skip to main content

Patterns to Genes

Project Goals 

To understand the heritability of physical patterns, for instance spots in leopards or stripes in zebras. The approach leverages computer vision methods to quantify similarity between patterns and to align those with known pedigrees to understand genetic relatedness and pattern similarity. 

Project Overview

Using ResNet50 features, we have achieved high purity levels when clustering egg images by their clutches. Further, using Gabor filter-based features we have achieved high levels of accuracy for classification by clutch. Using machine learning and AI we have been able to determine that there is  a very strong likelihood that there is a texture signal in the egg patterning.