Construction of apricot variety search engine based on deep learning

• DOI: 10.1016/j.hpj.2023.02.007
• Publication Date: 2023-02-18T01:43:19Z
• Authors: Chen Chen; Lin Wang; Huimin Liu; Jing Liu; Wanyu Xu; Mengzhen Huang; Ningning Gou; Chu Wang; Haikun Bai; Gengjie Jia; Tana Wuyun
• Keywords: Identification; Phenomics; Prunus armeniaca

Apricot has a long history of cultivation and many varieties types. The traditional variety identification methods are time-consuming labor-consuming, posing grand challenges to apricot resource management. Tool development in this regard will help researchers quickly identify information. This study photographed fruits outdoors indoors constructed dataset that can precisely classify the using U-net model (F-score: 99%), which helps obtain fruit's size, shape, color features. Meanwhile, search engine was constructed, from database according above Besides, mobile web application (ApricotView) developed, construction mode be also applied other fruit trees. Additionally, we have collected four difficult-to-identify seed datasets used VGG16 for training, with an accuracy 97%, provided important basis ApricotView. To address difficulties data collection bottlenecking phenomics research, developed first platform its kind (ApricotDIAP, http://apricotdiap.com/) accumulate, manage, publicize scientific apricot.