Medicinal Plant Species Detection using Deep Learning
K.C. Kavitha, Prashant Sharma, Shubham Gupta, R. V. S. Lalitha
Abstract
In the present era, the world is trending towards Automated Systems without human intervention with rapid development in Technology. One such area is medicinal plant species detection, where an expert identifies the medicinal leaf based on their botanical knowledge. We propose a deep learning model to detect the medicinal plant species based on its leaf image and advanced computer vision. This paper compares the Convolutional Neural Networks (CNN) variants viz., MobileNet, ResNet50, Inception v3, Xception, and DenseNet121 for Indian origin medicinal plant species detection. We have evaluated CNN variants to classify the medicinal leaf images and observed that the Inception v3 model outperforms all other conventional methods. Our proposed architecture adopts the Inception v3 model and the stochastic gradient descent technique during the training process for optimizing and achieving better results. Our experimental results show that the Inception v3 model achieved 95% accuracy in the Indian origin medicinal plant species classification