2021/10/29 by Abdollah Zakeri, Rohollah Hedayati, Zakeri, Abdollah +6
Agricultural and Biological Sciences · Chemistry · Earth and Planetary Sciences · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Remote Sensing and Land Use #Spectroscopy and Chemometric Analyses #Ziziphus Jujuba Studies and Applications
paper · pdf · doi:10.48550/arxiv.2111.00112
openalex publication_date 2021/10/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Jujube is a fruit mainly cultivated in India, China and Iran and has many\nhealth benefits. It is sold both fresh and dried. There are several factors in\njujube pricing such as weight, wrinkles and defections. Some jujube farmers\nsell their product all at once, without any proper sorting or classification,\nfor an average price. Our studies and experiences show that their profit can\nincrease significantly if their product is sold after the sorting process.\nThere are some traditional sorting methods for dried jujube fruit but they are\ncostly, time consuming and can be inaccurate due to human error. Nowadays,\ncomputer vision combined with machine learning methods, is used increasingly in\nfood industry for sorting and classification purposes and solve many of the\ntraditional sorting methods' problems. In this paper we are proposing a\ncomputer vision-based method for grading jujube fruits using machine learning\ntechniques which will take most of the important pricing factors into account\nand can be used to increase the profit of farmers. In this method we first\nacquire several images from different samples and then extract their visual\nfeatures such as color features, shape and size features, texture features,\ndefection and wrinkle features and then we select the most useful features\nusing feature selection algorithms like PCA and CFS. A feature vector is\nobtained for each sample and we use these vectors to train our classifiers to\nbe able to specify the corresponding pre-defined group for each of the samples.\nWe used different classifiers and training methods in order to obtain the best\nresult and by using decision tree we could reach 98.8% accuracy of the\nclassification.\n