Exploring agricultural image data through deep learning for visual information analysis : A study on Cabbage (Brassica Oleracea var. Capitata) farming

Abstract

This article outlines a methodology for developing and evaluating efficient image classification models using Convolutional Neural Networks (CNNs). It begins with meticulous network architecture design and training on a dataset comprising 9000 images of Cabbage, Weeds, and empty areas in cabbage plants, with the aim of achieving accurate image classification based on features and patterns. The paper conducts a detailed comparative analysis with established models such as AlexNet and ResNet, as well as modified versions of AlexNet and ResNet, employing performance metrics such as accuracy, precision, recall, and the F1 score. This comprehensive evaluation highlights the proficiency of the developed models and their relative effectiveness. Real-world datasets featuring fragmented agricultural plot images validate the models in practical scenarios, affirming their accuracy and reliability in agricultural data analysis. This article is helpful for the implementation of CNN in the automation of weeding in cabbage crops.