A Deep Neural Network for Multi-class Dry Beans Classification
Md. Mahadi Hasan, Muhammad Usama Islam, Muhammad Jafar Sadeq
Abstract
The technological explosion has paved the way for agriculture to flourish exponentially thus contributing to better yield of crops through the aid of machine learning, the Internet of things, mechanical systems in agriculture. In our research work, we have investigated various types of dry beans followed by a deep neural network based approach to classify the beans automatically. The results shows that our approach had an accuracy of 93.44%, and an F-1 score of 94.57%, with the dataset that consisted of 7 varieties of dry beans. Our results, which performed substantially better in comparison to traditional machine learning approaches aided us to devise further research scopes in the field of agricultural machine learning.