Machine Learning based Leaves Classifier using CNN and Reduced VGG Net Model

The term Machine Learning is broadly used in the last two decades. It makes much rapid progress in the area of machine vision. Just because of the arrival of Convolution Neural Networks the computer vision gets much better accuracy as compared to classical Machine Learning algorithms. The arrival of Neural Nets helps in complex classification and detection from the image. In this work, we investigate the VGG-Net model which was originally proposed by Oxford University. The VGG-Net architecture is very large and comprises a large space of memory and computational power and it has many layers. The reduced three layers are developed that comprise of
$5 \times 5$
convolution filters which also decrease computational power and memory consumption. Our model is software-based in which we used NVidia GPU for the training of our model that achieve a great accuracy in between 98% to 99% for different types of leaves classes. Our proposed model helps in the field of botany to classify different species of plants leaf and helps in the study of plants leaf. If there were no specialists in the field of botanists then our software-based model helps to classify which type of leaf is this.