Smart Module Design for Refrigerators based on Inception-V3 CNN Architecture

Abstract

The new era of advancement has started to attract a lot of eyes towards home automation and one of the most used items in our home is the refrigerator. After the launch of some smart refrigerators from well-established brands like Samsung, a new research area has opened up for developers worldwide as the smart fridges available in the market were aimed towards the higher class of individuals. Many developers and researchers have proposed and implemented several designs to build low cost smart refrigerator modules based on latest technologies like RFID, IoT, Image Processing and Vision, Artificial Intelligence and Machine Learning. This paper has discussed some of the previously built smart modules and the pros and cons of them. The paper also proposes an innovative, portable and cost-effective module design for making any ordinary refrigerator smart based on Inception-V3 CNN Model, Raspberry Pi 3B+ and an inbuilt camera that would help in scanning the products to be placed in the refrigerator. The module has been tested using standard datasets like FIDS30, Combined and Fruits360 and accuracy thus obtained is 89.7%, 85.8% and 94.9% respectively. Moreover keeping in mind the situation of Covid-19 Pandemic the module also comprises a UVC Disinfection box that would help in eliminating the risk of the virus entering our homes.