期刊论文

中文

中国粮油学报

1003-0174

2023

38

2

146-152

10.3969/j.issn.1003-0174.2023.02.022

湖北省自然科学基金项目（2016CBF470）； 湖北省教研项目（2021351）；

本校为第一机构

稻米是我国人民食用的主要粮食作物，因此稻米的品种识别与质量检测有着重要的意义。在以前，通常使用的是人工的方式用眼睛观察来识别不同品种的稻米，这种方法效率低下，而且容易受主观的影响导致识别出错，从而影响准确率。本文基于MATLAB平台设计一套稻米品种识别系统，首先使用图像处理将实际拍摄稻米图片进行目标检测和提取，后采用迁移学习的方式训练GoogLeNet深度神经网络模型对稻米图片进行识别，不仅如此该模型还能检测出每一粒稻米是否为整精米，且识别准确率可达到96.46%。在相同训练参数下对比VGG19训练结果，验证精度提高1.46%，调整数据量探究其对模型性能影响，模型准确率随着数...

Rice is the main food crop for Chinese people, so it is of great significance to identify rice varieties and detect rice quality. In the past, artificial methods were usually used to identify different varieties of rice by eye observation, which was inefficient and pron...MORE Rice is the main food crop for Chinese people, so it is of great significance to identify rice varieties and detect rice quality. In the past, artificial methods were usually used to identify different varieties of rice by eye observation, which was inefficient and prone to subjective influence, resulting in recognition ...