ABSTRACT
Accurate classification is an inevitable and indispensable need of Machine learning applications, where the decisions taken are crucial and never be reverted without consequences. Hence the importance of ensuring confidence in decision making is the need of every such application. We used convolutional neural network to predict the real flower images from their close challengers like, drawings, posters, ornamental replicas, dress models etc. A proper application of this CNN technique with Max pooling identifies the genuine image clearly out of near original and replicas to enhance the machine learning. In this work, we try to implement a more difficult task by categorizing the replicates as close as the originals and trained the network to predict the original flower images. We trained the network for 500 images and achieved the accuracy level of 90%. Experimental results shows that the proposed setup successfully predicted 90% flower and non flower images correctly.
