基于轻量化卷积神经网络的大米分选实验
Rice classification based on lightweight convolution neural network
在大米质量控制的问题上，传统的人工分选方法已经无法满足生产要求。通过对轻量化卷积神经网络在大米分类问题上的研究，检验了MobileNet、Xception、ShuffleNet三个轻量化卷积神经网络在碎米与垩白米分选问题上的效果，并与若干传统方法做了对比。结果证明轻量化卷积神经网络一方面能有效分选碎米与垩白米，另一方面能保证检测的实时性。
On the issue of rice quality control, traditional manual sorting methods have been unable to meet production requirements. Through the research on lightweight convolutional neural network in rice classification, the effect of three lightweight convolutional neural networks (MobileNet, Xception, and ShuffleNet) on the classification of broken rice and chalky rice was examinedand compared with several traditional methods. The results prove that the lightweight convolutional neural network can ensure real-time detection while effectively sorting broken rice and chalky rice.
董玉德、甘骐榕、徐道际
农业科学技术发展计算技术、计算机技术农作物
大米色选卷积神经网络轻量化
riceclassificationconvolution neural networklightweight
董玉德,甘骐榕,徐道际.基于轻量化卷积神经网络的大米分选实验[EB/OL].(2018-06-01)[2025-08-02].http://www.paper.edu.cn/releasepaper/content/201806-7.点此复制
评论