Optimization and Classification for Plant Leaf Detection Using Multi-Objective Genetic Algorithm and Random Forest

The agricultural sector in India, a developing nation, relies heavily on agriculture for its population's livelihood and food security. Plants are essential for sustenance, medicine, and various industries. This study is focused on the production and subsequent evaluation of an advanced plant disease classification method using Plant Village dataset. The proposed approach integrates texture feature extraction, multi-objective optimization, and Random Forest Classifier to enhance classification accuracy and efficiency. The NSGRF approach, a key component of the system, significantly outperforms in terms of precision, F-score, recall, and accuracy. In experiment survey the algorithm NSGRF improves 2% accuracy, 1 % precision and 2 % of recall and 1% of f-score as compared to other approaches. Author describes the process of the proposed model across various classes of Tomato, Potato and Pepper-Bell. They have successfully verified the performance of proposed model on a plant village dataset consisting of 15 different classes, including Tomato, Potato, and Bell-Pepper. Additionally, the study discussed their findings and contrasted this with the proposed model. The proposed NSGRF method exhibits superior performance in comparison to alternative methods, achieving the utmost 99.7% of accuracy with an error rate of 0.3%, 99.72% of precision, 99.8% of recall, and 99.78% of F-score. The study's findings indicate that this integrated approach is not only more accurate but also offers faster inference times and improved interpretability, making it a promising tool for automated plant disease detection within the domain of agricultural technology.