Deep Learning Based Classification For Paddy Pests&Diseases Recognition

Paddy is a staple food in Indonesia,the availability of paddy is really important to sustain life.Paddy pests and diseases are one major factor that is threatening the availability of paddy.However,the identification of paddy pests and diseases itself remain as a challenge for paddy farmer in Indonesia.Because of the lack of understanding and training.Also the lack of an extension worker to assist the farmer to help them identify pests and diseases.Increasing smartphone usage in Indonesia,at the same time of deep learning advancement,create an opportunity to solve this challenge.Collecting 4,270 dataset images from four Asian languages that submitted into the search engine.CaffeNet model uses this dataset to train the model and producing 86% accuracy.ResNet model that training using the same dataset achieving 87% accuracy.The implementation prototype is given after model experimentation,to prove the feasibility of the model to solve a real-world problem.