Chicken Egg Gender is Identified Using Image Analysis Techniques in a Non-Invasive Manner

Abstract

Approximately 7 billion new hatch male chicks are being destroyed annually due to the fact that only females are retained for egg production. Not only is this practice unethical, but it also results in tremendous monetary loss. While today's technologies are still limited in terms of precision, it is possible to determine the gender of chicks either pre-hatch or post-hatch, with pre-hatch determination being more advantageous. Looking at the Shape Index (SI),or the ratio of the short and long diameters of an egg, is one method. Although SI differs according to egg type, there exists a high correlation between SI and chick sex (r = 0.78). Although chickens' discrimination is not as good as ducks', SI is a point of reference for gender prediction. We included seven more morphological parameters in our analysis in the present study: mass, Modality, volume, eccentricity, short axis, and long axis. By applying a pre-trained Tensor Flow model and earlier research predictions, we were successful in classifying with 71% accuracy, with our predictions very close to 80% accuracy. If employed on a large scale, such a strategy can rescue the killing of nearly 5.65 billion chicks every year and curb egg waste. As a result, egg morphology, especially the shape index, presents a possible non-invasive method of sexing chicks before hatching.