Macroscopic pest detection system for aromatic herbs in post-harvest stage using computer vision and neural networks techniques

Abstract

Worldwide consumption of aromatic herbs has increased in recent years, which has led to an increase in the areas of cultivation and exports of these plants in producing countries. Within the production chain of these herbs, phytosanitary control is a requirement for the quality assurance and exports of the product. Pests cause damage in the plants, not only in the crop but also after the harvest. According to the regulatory authorities, who conduct inspections for pests, bad phytosanitary practices lead to the return of the product from the port to its origin or destruction in the country of destination. This results in an economic loss to the producer and exporter of the plant. One of the aromatic herbs with the highest level of production and export in Colombian market is thyme (Thymus). This work presents a support system for detecting macroscopic pests in thyme using computer vision techniques and neural networks. The main contribution of this work focuses on product classification between free and not free for pest in the postharvest process.