Numerical modeling of image discriminability for home storage and organization system on a smart device

Abstract

In Home storage and organization system on a smart device, thumbnail pictures (Tag Image) of daily-use objects are often used. Discriminability of Tag Image is important to realize superior usability. In this paper, we have tried to construct a numerical model of Tag Image's discriminabillity. The proposed model is based on simple linear regression from popular image features and their statistics. In addition, web-based data input system has also been developed to collect training data efficiently. Consequently, the input system has acquired a substantial number of data and a numerical model has been constructed. The constructed model has substantially good but not perfect performance.