Leaf Species Classification Using Teeth Features

Abstract

Plant recognition is a vital and challenging task. Leaf recognition plays a vital role in plant recognition and its key issue lies in whether selected features are solid and have a good potentiality to prejudice between different kinds of leaves. Leaf teeth are one of the most essential and intricate aspects of the leaf that are employed in automatic plant identification systems to classify and identify plant species. Automatic species identification has a huge amount of advantages over traditional species identification. Presently, most plant automatic identification strategies focus on shape, venation and texture, a feature unremarkably utilized in ancient species identification that is unnoticed. Different species of leaves have distinct characteristics that aid in the classification of certain plant species. These features aid botanists in more precisely recognizing major species of plants from leaf images. One of the most essential and complex aspects of leaf in plant species is the teeth. The leaf images are pre-processed in this study, and the segmentation is done using Canny edge detection. The features are extracted using a feature extraction model and CNN model namely fine-tuned VGG16. These features are then classified using the CNN model and acquired an average accuracy of 95.45 percent.