Detection and classification of chicken diseases has been a longstanding concern for farmers. Relying solely on expert opinions could be a costly option. Moreover, there are certain chicken diseases that spread quickly, and necessary procedures are immediately needed to carryon. On the other hand, the lack of research activity or low availability of data on this topic leads to many misunderstandings and challenges so as to correctly detect a certain type of chicken disease. The existing techniques show low accuracy on disease classification, which leads to a complex engineering problem and critical analysis is required for sustainable development. In this paper, we develop a novel hybrid deep neural network, namely ChickenNet21, which combines a Convolutional Neural Network (CNN) and a fine-tuned Visual Geometry Group (VGG) model to address this challenge. Using a large dataset of 6,812 farm-labeled fecal images, we assess the effectiveness of ChickenNet21 and it achieves 98.83% accuracy in identifying chicken diseases.