The Graded CNN Technique to Identify Type of Food as The Preliminary Stages to Handle the Issues of Image Content Abundant

Abstract

In this social media era, the marketing promotion is shifting from text to multimedia content. There is needed the new techniques and algorithm to deal with unrequested content posted to our social media page. The paper presents the study result of the application of Deep Learning method to identify the type of food i.e. junk-food or healthy food. In the experiment we explore some various layers of the Convolution Neural Network in classifying the type of food. We proposed another configuration stated as graded CNN. The results of the experiment show that our graded CNN technique proposed is outperform compared to the other trivial CNN configuration. Both of performance parameters, i.e. accuracy and time to process, confirm that our graded CNN technique is feasible to be considered as the powerful CNN variant in image classification domain. The average accuracy of the graded CNN is 9 % better than the common CNN, whereas the time to process is 400% more efficient.