A study on egg sex-related associations based on PP-HGNet and morphological parameters

Area of Science:

• Poultry Science
• Animal Breeding
• Agricultural Technology

Background:

• Accurate pre-hatch egg sex determination is crucial in the poultry industry to manage male chick culling.
• Previous research on Egg Shape Index (EI) for sex identification yielded inconsistent results, often using limited, non-commercial datasets.
• A need exists for industry-relevant studies with larger sample sizes and commercial breeds.

Purpose of the Study:

• To systematically analyze the potential of external egg morphology for sex identification in commercial poultry breeds.
• To evaluate the efficacy of statistical analysis and a deep neural network (PP-HGNet) for sex determination using egg features.
• To provide a robust, industry-relevant dataset and analysis to clarify the utility of EI for sex identification.

Main Methods:

• Analysis of 1008 eggs from two commercial layer breeds (Xiaoyou 818 and Datu2).
• Evaluation of external morphological features using statistical analysis.
• Application of a deep neural network (PP-HGNet) for predictive modeling.
• Comparison of results across breeds and methods.

Main Results:

• No statistically significant differences were found between male and female eggs based on morphological features.
• The deep neural network (PP-HGNet) achieved predictive accuracy comparable to random chance.
• The study demonstrated the limitations of EI and morphological traits for reliable sex identification.

Conclusions:

• Egg Shape Index and related external morphological features are insufficient for accurate pre-hatch sex identification in commercial poultry.
• The large-scale, multi-breed approach confirms the ineffectiveness of these methods.
• Future research may need to explore alternative, non-invasive sexing technologies for the poultry industry.