Image Classification for Retail

Image classification, an area of Deep learning is used to classify images according to certain class it belongs to. Classification technique is used in many domains but its use in retail sector has been limited due to very complex problems and limitation to the quality of the images. Image captured and its quality, used for training image classifier play an crucial role for deciding potential correctness of classifier. In Retail market certain policy is observed for arranging items(planogram) so that sale of those items can be improved. To check for implementation of these policy large human resources are deployed by the company. Using precise image classification we can solve this complex problem of planogram compliance for retail use case. In this paper we propose an hybrid approach of image classification along with object detection for better image classification. We also added BRISQUE image quality assessment in our framework for better image classification. Better sale can be achieved and decisions can be taken prior by studying the results of classifier.