PROJECT TITLE :

Deep Learning for Plant Species Classification Using Leaf Vein Morphometric

ABSTRACT:

Botanists and laypeople alike could benefit from a system that automatically identifies plant species. Deep learning is effective for feature extraction since it provides more detailed information about images. D-Leaf, a new CNN-based approach, was proposed in this study. Three distinct Convolutional Neural Network (CNN) models, pre-trained AlexNet, fine-tuned AlexNet, and D-Leaf, were used to pre-process the leaf images and extract the features. Five machine learning algorithms were used to classify these features: Support Vector Machine (SVM), Artificial Neural Network (ANN), k-Nearest-Neighbor (k-NN), Nave-Bayes (NB), and CNN. For benchmarking, a traditional morphometric approach based on Sobel segmented veins was used to calculate morphological measurements. In comparison to AlexNet (93.26 percent) and fine-tuned AlexNet (95.54 percent) models, the D-Leaf model obtained a testing accuracy of 94.88 percent. Furthermore, CNN models outperformed standard morphometric assessments (66.55 percent). The ANN classifier is found to fit the characteristics derived from the CNN well.