Early Detection of Sunflower Leaf Diseases Using Image-Based Deep Learning Methods

Sunflower is a crop type that has high economic value and is also used for ornamental purposes. However, various diseases seen on sunflower leaves can disrupt production and it is difficult for growers to identify these diseases with traditional approaches. Therefore, the need for image-based artificial intelligence approaches that can automatically identify diseases seen on leaves has arisen. In this study, a system that can detect diseases seen on sunflower leaves, both image-based and artificial intelligence-supported, has been developed. The study consists of four stages. In the first stage, a publicly available dataset was used, and additional data was collected by us. In the second stage, image processing was performed. In the third stage, CNN (Convolutional Neural Network), ViT (Vision Transformer) and CNN-ViT models were designed. In the last stage, the performances of these models were evaluated, and their success was determined by accuracy, recall, precision, F1-score, Cohen Kappa and Hamming loss metrics. Experimental results revealed that the hybrid approach used in the study was more effective than traditional deep learning models.