CNN Based Object Classification

Abstract

Remarkable advancement has been made in image recognition, essentially because of the availability of large scale annotated datasets (i.e. ImageNet) and the recovery of deep convolutional neural systems (CNN). CNNs enable learning data driven, highly represen- tative. However, obtaining datasets as comprehensively annotated as ImageNet in the herbal plant domain remains a challenge. There are currently three major techniques that successfully employ CNNs to herbal plant image classification: training the pretrained model through transfer learning ,training the CNN from scratch, using off-the-shelf pre-trained CNN features. First explore and evaluate different architectures of CNN that is AlexNet, VGG16, VGG19 and GoogLeNet. We report the five-fold cross-validation on herbal plant classifi- cation and predict the accuracy on LRD (Liquorice, Rhubarb, Dhatura) categories. The accuracy of the classification of herbal plants will be depends on the input size of the image.