Mango Leaves Recognition Using Deep Belief Network with MFO and Multi-feature Fusion

In automatic plant classification, plant identification based on digital leaf images is a challenging task. This paper proposes a Moth-Flame Optimization (MFO) based Deep Belief Network (DBN) method for plant leaf recognition. Initially, a combination of texture and shape features is applied for extracting features from the preprocessed image. Further, for leaf classification, the MFO optimizes the DBN parameters to minimize error and are used as classifiers. The classifier has been applied to five different sets of mango leaf images and achieved an accuracy of 98.5%. The experimental result indicates that it is feasible to automatically classify plants by using multi-feature extraction of plant leaf images in combination with MFO-based DBN. © 2020, Springer Nature Singapore Pte Ltd.