Advanced Chicken Breed Identification Using Transfer Learning Techniques with the VGG16 Convolutional Neural Network Architecture

This study proposes a deep learning-based classification system to identify chicken breeds from image data. A dataset of 2,400 labeled images representing twelve distinct chicken breeds was collected and divided into training, validation, and testing sets. The system employs transfer learning by integrating the Mobile VGG16 convolutional neural network as the feature extraction backbone. The extracted features were then passed through custom classification layers to differentiate among the breeds. The model was trained using 1,800 images, validated with 300 images, and evaluated on a separate test set of 300 images. During testing, the model achieved an accuracy of 81% and a categorical cross-entropy loss of 0.378. These results indicate that the model can effectively recognize subtle visual distinctions between similar-looking chicken breeds. The system demonstrates practical potential for applications in poultry farming, biodiversity documentation, and automated livestock management. The findings confirm that deep convolutional neural networks, particularly VGG16 in a transfer learning setup, are capable of performing fine-grained classification tasks in real-world scenarios. The proposed method provides a reliable and scalable solution for automatic chicken breed identification based on image input.