Using CNN for detection of diseases

Abstract

Since economic booming, people start to care more about their health including dietary structure and nutrition. Scientist announced that fish and chicken are the best meat among others based on high nutrition level and low fat percentage. However, It is proved that many diseases especially H7N9 are related to poultry farming. Those diseases not only lethal to chicken but also has certain chance to evolve to human transmittable illness. Unfortunately, industrialized large scale poultry farming increased the morbidity of the diseases which is enough for people to take concern of. After all this, the application of genetic engineering technology to poultry industry focused on increase productivity, which sadly reduce poultry healthiness. These lead to periodic poultry pandemic blast. This is endless periodic nightmare for poultry farming. Theoretically, Using image recognizing from artificial intelligent to detect diseases at first sign is possible with current research level. This paper will compare and record the changes of appearances of five common diseases and try to design an AI model to detect it. The main results are as follow: (Use convenlutional neutral network to judge whether the chickens are sick by checking the cockscomb, skin, feather and waste. Keep training the computer to correct the weight to let the accurate rate above 98%. Extract the data of the images by digitizing, and continually recognize the diseases through the help of the vets to realize back propagation to improve the accurate rate.)