MAGIC: Mobile App for Gender Identification of Chick from Vocalization Data Analysis

Efficient gender identification of day-old chicks is critical to modern poultry management, enabling streamlined breeding strategies. This study presents a comprehensive experimental investigation into gender detection using vocalization data gathered from four distinct day-old chick breeds: Cornish (CN), Gramapriya (GP), Vanaraja (VR), and White Leghorn (WL), sourced from the ICAR-DPR (Indian Council of Agricultural Re-search-Directorate of Poultry Research) hatchery. The vocalization data undergo rigorous preprocessing, including voice activity detection (VAD) and segmentation, to facilitate the ensuing experimental analysis. This study delves into the efficacy of four distinct feature sets—Mel-frequency Cepstral Coefficients (MFCCs), chroma features, Mel spectrograms, and hybrid features—as representations for the intricate vocalization patterns. Our central goal is to achieve precise gender classification, employing a diverse range of classifiers such as support vector machines (SVM), multilayer perceptron (MLP), logistic regression (LR), decision tree (DT), random forest (RF), gradient boosting machines (GBM), Gaussian Naive Bayes (GNB), K-nearest neighbors (KNN), X gradient boost (XGB), AdaBoost (ADB), and LightGBM (LGB). To determine the optimal classifier, we implement a meticulous polling mechanism that aggregates the outcomes of the experimental evaluations. The empirical findings distinctly highlight the superiority of hybrid features—an amalgamation of the explored feature sets—surpassing the performance of individual features. Our experiments demonstrate considerable progress in gender detection accuracy for day-old chicks. These findings hold substantial promise for refining breeding program management within the dynamic landscape of the poultry industry. Our work culminates in the prospect of integrating these insights into a mobile application (MAGIC: Mobile App for Gender Identification of Chick) that facilitates on-site gender identification, enhancing the efficiency of poultry breeding endeavors.