A Review on Deep Learning and Transfer Learning Models: Grape Leaf Disease Detection

Grapes are the most significant fruit crops which contain several vitamins and minerals such as vitamins K, and B, as well as copper. Grape plants can be greatly affected by different diseases including grape esca, isariopsis leaf spot, black rot, powdery mildew or downy mildew, and so on. These diseases impact the entire grape's quality and yield. Early detection of grape leaf disease is essential to reduce economic costs and disease progression. However, various existing approaches pose certain challenges including computational cost, real-time detection, illumination, and occlusion as well as misclassification. This survey analyzes recent works using the machine learning, deep learning, and transfer learning techniques for grape leaf disease diagnosis in the form of dataset availability, disease severity estimation, and computational resources. Among other techniques, deep learning shows superior performance results and a better capability to extract features. Additionally, this research analyzes various performance metrics and the limitations of existing approaches.