Intelligent Computational Methods for Silkworm Disease Detection in Sericulture

Abstract

The cultivation of silkworms (Bombyx mori L.) to produce silk fabric called sericulture is an essential agroindustry that provides for many residents in rural territories of Asia, Europe, and South America. Despite its economic significance, the industry suffers from serious productivity problems because of the infectious diseases including Pebrine, Grasserie, Flacherie, and Muscardine, causing the 15-25% yield losses per year globally. The traditional approach for detecting the diseases in question is visual inspection, yet it might be rather subjective resulting in delayed diagnosis and inconsistent accuracy levels. Moreover, it cannot be implemented on a large scale.

This paper aims at analyzing the literature about the application of the AI and computational approaches in the automatic identification of diseases of silkworms. The review will cover 60 papers published from 2012 to 2025 and reveal the evolution of the technologies from classical image processing and machine learning to the latest deep learning algorithms including convolutional neural networks, transfer learning, and lightweight solutions for practical implementation. Modern AI techniques guarantee highly accurate disease detection (above 95%) in controlled environment. However, some challenges are identified including the inconsistent data standards, unreliability in practice and integration with pest management system. Our key findings include research gaps in the form of lack of benchmark data, insufficient validation in practice, and limited understanding of the restrictions of small farmers. In addition, we present a research protocol for improving comparisons among studies and future directions including multi-modal sensing, Internet of Things, and explainable AI in practice. This review is intended to provide fundamental information for developing precision sericulture using technology from different areas.