Hybrid Optimization of Coccidiosis Chicken Disease Prediction, Detection and Prevention Using Deep Learning Frameworks
Keywords: Coccidiosis, deep learning, model
TL;DR: Using Artificial intelligence to make an automatic vaccination sysytem for poultry
Abstract: Poultry farming is one of the thriving businesses in Kenya, therefore, playing a crucial role in the economy and the food value chain. Most of the farmers are small scale while a good number practice large-scale farming. Egg-laying birds are the most preferred as a result of the high profits gained from egg sales. However, various stresses including disease outbreaks have greatly caused loss due to late detection and lack of systems to predict the diseases before the infections. Coccidiosis has been one of the most prevalent and highly contagious poultry diseases. As such, there is a need to address the challenges, by employing emerging technologies in a built environment. In this study, deep learning models were deployed in the TensorFlow framework to detect the onset of the disease. Due to the fast spread of the disease, an automated vaccination system was used to protect healthy birds from attaining the disease, and to adopt a robust prevention system, a disease prediction framework was deployed to alert farmers to adapt to better mechanisms. Different deep learning models were deployed and tested and their accuracies were compared to get a fully efficient model. A Convolutional Neural Network (CNN) ResNet50 model showed the highest accuracy of 96\% through the transfer learning technique. The deployed automatic vaccination system revealed high efficiency in releasing the right dose amounts after the disease is detected. Therefore, the Integration of engineering technologies to foster automated systems will not only ensure food security in poultry farming but also open up more industries to build these systems for the long-term benefit of farmers.
Submission Category: Machine learning algorithms
Submission Number: 40
Loading