SIFT-Based Prickly Plant Identification System for Visually Impaired People

A machine learning model to classify a plant as prickly or non-prickly is presented in this work. The work mainly addresses the issue of injuries sustained by visually impaired workers during farm activities such as harvesting and reaping. These tasks frequently entail handling prickly plants, which can result in severe injuries, especially for persons who are visually impaired and may not be able to quickly identify possible dangers. By combining sensors and algorithms to identify and warn workers of potential hazards in their immediate surroundings, the model presented in this paper offers a solution to this issue. A novel algorithm is implemented to tune the hyper parameter based on k-nearest neighbor. The quantitative features such as leaf size, nearest neighbor, and weights contribute toward the highest accuracy. Three supervised machine learning classifiers Random Forest, Decision Tree, and KNN are utilized. Among all three, highest accuracy of 90.91% is achieved using Random Forest.