摘要
图像识别作为深度学习领域内的一项重要应用,水果图像的分类识别在智慧农业以及采摘机器人等方面具有重要应用。针对以往传统图像分类算法存在泛化能力差、准确率不高等问题,提出一种在TensorFlow框架下基于深度学习和迁移学习的水果图像分类算法。该算法采用Inception-V3的部分模型结构对水果图像数据进行特征提取,采用Softmax分类器对图像特征进行分类,并通过迁移学习方式进行训练得到迁移训练模型。测试结果表明,该算法与传统水果分类算法对比,具有较高识别准确率。
Image recognition is an important application in the field of deep learning.The classification and recognition of fruit images has important applications in smart agriculture and picking robots.Inorder to solve the problems of poor generalization ability and low accuracy in traditional image classification algorithms,a fruit image classification algorithm based on deep learning and migration learning under TensorFlow framework is proposed.The partial model structure of Inception-V3 is used to extract the feature image of fruit image,and Softmax classifier is used to classify the image features,which trained by migration learning to obtain the migration training model.The test results show that the algorithm has higher recognition accuracy than the traditional fruit classification algorithm.
作者
廉小亲
成开元
安飒
吴叶兰
关文洋
LIAN Xiao-qin;CHENG Kai-yuan;AN Sa;WU Ye-lan;GUAN Wen-yang(Beijing Key Laboratory of Big Data Technology for Food Safety,School of Computer and Infomation Engineering,Beijing Technology and Business University,Beijing 100048,China)
出处
《测控技术》
2019年第6期15-18,共4页
Measurement & Control Technology
基金
北京高等学校教育教学改革项目(2015-ms146)
科技创新服务能力建设(PXM2018_014213_000033)
北京工商大学研究生培养-研究生教育质量提升计划项目(19008001491)
