CASHEW NUT IDENTIFICATION IN PACKAGING AND QUALITY INSPECTION LINES BASED ON CONVOLUTIONAL NEURAL NETWORK

Abstract

This paper presents an algorithm to identify and classify cashew nuts in the
packaging and quality inspection lines based on the YOLOv3 convolutional neural
network (version 3). In addition, we describe here a dataset for classifying good,
broken, and shelled cashews. The experimental results show that the proposed
algorithm can identify cashew nuts with an mean average precision (mAP) over
our dataset is of 97.34%.