Leaf Disease Detection Using Machine Learning and Deep Learning

Abstract

This thesis presents algorithms that not only classify images of healthy and infected leaves but also deliver results with low execution times. The novel 1-norm twin random vector functional link (UTRVFL1norm) networks based on Universum data and 1-norm twin random vector functional link (TRVFL1norm) networks have been proposed, which provide superior performance on both benchmark and artificially created datasets compared to conventional models. UTRVFL1norm and TRVFL1norm effectively fulfill their objectives by appropriately classifying benchmark datasets and small-sized artificial datasets. Further, an innovative algorithm called 1-norm entropy-based fuzzy twin random vector functional link (EFTRVFL1norm) networks has been developed, which effectively classifies imbalanced binary datasets using the concept of entropy-based fuzziness. The comparative analysis clearly demonstrated that the proposed model outperformed the other models. newline newlineAlong with this affinity-based fuzzy random vector functional link (ACFRVFL) networks have been designed to address the challenges posed by noisy and imbalanced datasets. While many algorithms are designed to classify either noisy datasets or imbalanced binary datasets, it is crucial to develop algorithms that can successfully handle both. This research also evaluates the performance of this model on noise-affected artificial leaf datasets and imbalanced datasets. The findings of this thesis demonstrate that the proposed schemes/models outperform existing ones, achieving significantly better performance. newline newline