A Systematic Review of Machine and Deep Learning Techniques for Cassava Disease Detection: Trends, Challenges, and Pathways to Real-World Implementation

Abstract: Cassava, a major staple crop in sub-Saharan Africa, faces significant threats from leaf diseases such as Cassava Mosaic Disease (CMD) and Cassava Brown Streak Disease (CBSD). Early and accurate detection is essential for timely intervention. While artificial intelligence (AI), including Machine Learning (ML) and Deep Learning (DL), has gained traction in agricultural disease detection, inconsistencies across studies limit practical applicability. This systematic review aims to synthesize peer-reviewed research from 2020 to 2025 on AI applications for cassava disease detection using image-based methods. The review investigates model types, dataset origins, evaluation metrics, cross- comparative evaluation, and deployment readiness. A structured search was conducted across databases (IEEE Xplore, Scopus, SpringerLink, ScienceDirect, and Google Scholar), applying PRISMA guidelines. A total of 100 eligible studies were selected based on inclusion criteria. Metadata extraction covered AI methodology, dataset characteristics, performance evaluation, deployment features, and study limitations. Deep Learning dominated (65%) with CNN variants widely used, followed by ML (20%) and hybrid models (10%). PlantVillage was the most common dataset (64%), while field data accounted for only 21%. Accuracy was the most reported metric (95%), but only 20% used multi-metric evaluations. Cross-validation and cross-comparative evaluations were inconsistent. Few studies addressed deployment readiness or real-world applicability. Although AI methods have demonstrated high potential in cassava disease detection, significant gaps remain in dataset diversity, model evaluation rigor, deployment design, and ethical considerations. This review provides a road map for future research, emphasizing the need for standardized benchmarking, mobile-friendly models, and real-world field validation. Keywords: Cassava, Deep Learning, Machine Learning, Disease Detection, Agriculture. Title: A Systematic Review of Machine and Deep Learning Techniques for Cassava Disease Detection: Trends, Challenges, and Pathways to Real-World Implementation Author: Chika Norah John, Obi Chukwuemeka Nwokonkwo, Anthony Ifeanyi Otuonye, Ikechukwu Ignatius Ayogu, Innocent Harvey Ajunwa International Journal of Novel Research in Computer Science and Software Engineering ISSN 2394-7314 Vol. 12, Issue 2, May 2025 - August 2025 Page No: 49-64 Novelty Journals Website: www.noveltyjournals.com Published Date: 04-August-2025 DOI: doi.org Paper Download Link (Source) noveltyjournals.com