Disease Identification and Categorization in Pigeon Pea Leaves Using LBP and HOG Features
摘要
Pigeon pea (Cajanus cajan), a vital leguminous crop, is an abundant source of essential amino acids, and dietary fiber, making it a vital component of diets in many developing countries. In India, it is employed in intercrop and rotation systems with various cereal crops. However, pigeon pea plants are susceptible to several disorders that can severely impact harvest and quality. Leaf spot and Sterilic mosaic diseases are among several common pigeon pea leaf illnesses. In this paper, identification and classification of these disorders is presented. Grab cut strategy is used to eradicate undesired portion of the leaf photo. Features are extracted using local binary patterns (LBP) and histogram of oriented gradients (HOG) approaches. Dimensionality reduction on features is done using principal components analysis (PCA). Machine-based classifiers, namely, random forest, HistGradientBoost, K-nearest neighbor, and support vector machine are used to classify the plant diseases. To validate the results accuracy, precision, f1-score, and recall parameters are computed.