Improved Detection of Fusarium Head Blight in Wheat Ears Through SOLO Instance Segmentation

Wheat Fusarium Head Blight is a destructive fungal disease that affects wheat crops worldwide. It is caused by the fungus Fusarium graminearum, which infects the flowering heads of wheat plants during periods of high humidity and rainfall. The disease can lead to significant yield losses and reduced grain quality, making it a major concern for wheat farmers. In order to effectively manage and control Wheat Fusarium Head Blight, accurate detection and diagnosis of the disease is crucial. Traditional manual inspection techniques for detecting Wheat Fusarium Head Blight have limitations, including the need for expert knowledge, inaccuracies in diagnosis, and are labour-intensive. In light of the challenges posed by FHB, there is a pressing need to explore innovative research for FHB management. To address these limitations, researchers have developed a novel approach using deep learning and image segmentation techniques. This research paper proposes the use of SOLO (Segmenting Objects by Location), a one-shot object detection model, to enhance the precision and speed of lesion segmentation in wheat leaves affected by Wheat Fusarium Head Blight. Researchers capture detailed images of infected and healthy wheat kernels and these images can then be analysed using texture analysis techniques to extract key features that differentiate infected kernels from healthy ones. The proposed approach has achieved an impressive accuracyof 99.16%, signifying a significant scientific breakthrough. This accomplishment not only provides valuable assistance to researchers but also equips farmers with an advanced technological arsenal to combat Fusarium Head Blight.