Work

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

Year: 2026

Type: article

Abstract: In the layer industry, a large number of male chicks produced during incubation are economically unviable, as they neither lay eggs nor possess the rapid growth traits of broilers. Therefore, achieving accurate sex identification before hatching is of substantial practical importance. Existing methods, including invasive molecular assays, hormonal analysis, and non-invasive spectroscopic or machine-vision approaches, face limitations such as invasiveness, high equipment costs, delayed detection, or limited predictive accuracy. In contrast, morphological measurements, including the egg shape index (EI) and other derived parameters, offer a non-destructive, low-cost, and easily obtainable alternative. In this study, 1008 eggs from two breeds (Xiaoyou 818 and Datu2) were analyzed by measuring seven morphological parameters: length (L), width (W), EI, L+EI, W+EI, curvature (C), and C+EI. Statistical analyses and deep neural network PP-HGNet classification were conducted to assess their potential for predicting sex. The results showed no significant differences (P > 0.05) between male and female groups, and the model’s predictive accuracy was close to random. These findings indicate that EI and derived morphological parameters are unreliable for early sex prediction, highlighting the need to combine additional non-invasive techniques to achieve efficient and accurate sex determination in industrial incubation.

Source: Veterinary and Animal Science

DOI: 10.1016/j.vas.2026.100629