Random search Based Hyperparamerter tunned VGG16 Architecture for Poultry Breed Image Classification

Animal breed classification using deep learning algorithms is increasingly important due to its significant implications for agriculture, biodiversity conservation, and the poultry industry. Accurate breed classification aids in breed identification, disease management, and genetic research. This research aims to evaluate the performance and efficiency of the VGG16 architecture with hyperparameters optimized through Random Search for image classification tasks. Specifically, we compare the traditional CNN with the VGG16 architecture in poultry breed classification. Through rigorous evaluation, including various metrics, training efficiency analysis, and visualization techniques, we investigate the effectiveness of VGG16 with Random Search optimization. Our findings highlight the relative strengths and weaknesses of the proposed methodology, contributing to a deeper understanding of architecture in classifying different hen breeds. Using a dataset of 1010 hen breed images, we found that VGG16 with Random Search-based hyperparameter optimization outperformed the traditional CNN, achieving a 16% higher testing accuracy.