A Comprehensive Review of Soft Computational Methods for Silkworm Disease Detection: Advances, Challenges, and Future Directions

Abstract

Silkworm disease detection is a critical aspect of sericulture that directly impacts silk production quality and economic sustainability. This comprehensive review presents a systematic review of soft computational methods employed for silkworm disease detection, encompassing machine learning, deep learning, and image processing techniques. We analyze 60 research papers from prominent databases including IEEE, Springer, ScienceDirect, and Scopus, focusing on methodologies for detecting major silkworm diseases such as Pebrine, Grasserie, Flacherie, and Muscardine. Our analysis reveals significant advancements in convolutional neural networks (CNNs), transfer learning approaches, and hybrid ensemble methods achieving detection accuracies exceeding 95%. We provide detailed comparative analyses of plant disease detection methods and their applicability to silkworm disease identification, present comprehensive taxonomy of soft computing techniques, and identify key challenges including dataset limitations, real-time detection requirements, and deployment constraints. Furthermore, we propose future research directions emphasizing lightweight models for edge deployment, multi-disease classification systems, integration of Internet of Things (IoT) technologies for smart sericulture, and federated learning approaches for collaborative research. This review serves as a foundational resource for researchers and practitioners in computational sericulture, offering insights into current state-of-the-art techniques and emerging opportunities in automated silkworm health monitoring.