A Systematic Review of Recent Machine Learning Techniques for Plant Disease Identification and Classification

Plant disease management in agriculture science is the primary concern for every country, as the food demand is increasing fast due to an increase in population. Furthermore, modern technology has improved the efficacy and precision of disease detection in plants and animals. Plant disease identification using various image processing approaches has recently been employed on a big scale to help farmers monitor their plantation areas. Based on the perpetuation and spread, the diseases can be floral, foliar, and soilborne. Grain production is typically affected by foliar diseases, which reduce photosynthetic area, duration, and function. Soil-borne conditions include vascular wilt, root rot, and damping-off; and can exhibit symptoms such as wilting of foliage, root decay, and sudden death. This paper highlights the significant issues and challenges for leaf disease classification. A comparative study of different methods based on the agricultural product, methodology, efficiency, advantages, and disadvantages is also included. The review study analyzes the most frequently used machine learning algorithms in the last five to seven years, revealing that Support Vector Machine (SVM) has been extensively used for disease classification. An analysis of specific Techniques (Feature Extraction plus machine learning-based Classification algorithm) and their associated accuracy is also performed, demonstrating that (ORB) features combined with Linear SVM provide the highest accuracy of 99.98%.