A Durian Variety Identifier Using Canny Edge and CNN

Abstract

Durian belongs to the genus Durio and a native fruit in tropical countries in the regions of Southeast Asia such as Malaysia, Indonesia, Thailand, and the Philippines. Durian can be easily recognized by its spiky husk and pungent aroma. There are plenty of varieties of Durian which are hard to differentiate by someone who is even an expert of the said fruit. This study is conducted to utilize Raspberry Pi. It will be used to implement image processing algorithms and to create a portable device to identify the fruit’s variety. Image Processing has been used in today’s modern technology like in medicine, media, or the like, and devices are portable for accessibility. The researcher has considered the usefulness of this study and to further understand and acquire useful information of the portable device that will be created to identify Durian variants. Due to the multiple numbers of Durian varieties, the researcher has to determine 6 variants. As the study focused on the unknown variant, the researcher will use a guide on the datasets that will be needed for the training and testing phase of the software. The whole device would then be inspected and would document the accuracy of each variant. The data that the researcher will need to determine the quality and the accuracy of the software, further training and more testing will be needed to get the required acceptable accuracy percentage.