This research paper probes into the vital importance of poultry farming, specifically focusing on hens, which play a vibrant role in meeting the global demand for both eggs and meat. Identifying hen breeds and recognizing diseases pose significant challenges in poultry management, necessitating innovative solutions to enhance the efficiency of farming practices. The experimental efforts of this study were centered around classifying ten distinct hen breeds and recognizing four prevalent hen diseases through the implementation of an ensemble method. Utilizing a stacking-based ensemble approach, we achieved remarkable success, achieving a test accuracy of 99.94% for both hen breeds and 99.01% for disease classification based on feces images. In this study, we employed the self-collected dataset named ‘GalliformeSpectra’ for hen breed recognition, alongside a publicly accessible dataset of feces images to identify diseases. Additionally, to facilitate practical application, we have developed a smartphone application seamlessly incorporating our model, enabling real-time hen breed and disease classification. The findings of this study represent a groundbreaking accomplishment in the realm of hen breed classification using machine learning, distinguishing this study as both state-of-the-art and pioneering. By addressing critical challenges in poultry farming, this research contributes not only to academic progress but also provides practical solutions to enhance efficiency and sustainability in the poultry industry resulting in ease the farmers to be able to plan their farming business efficiently and to take measures in the correct time in case of diseases outbreak thus contributing to farmers, communities, and researchers.
Read full abstract