数字农科院2.0

Non-destructive viability detection of naturally-aged large-sample rice seeds using hyperspectral imaging and deep learning

文献类型： 外文期刊

作者： Hubo Xu;Fei Li;Ziqiang Wang;Wei Zhang;Hui Li;Xingyu Xia;Chunmei Li;Xia Xin

作者机构：

关键词： Cross-variety/germplasm;Hyperspectral imaging;Rice seed;Transformer;Viability detection

期刊名称： Journal of Agriculture and Food Research

ISSN： 2666-1543

年卷期： 2025 年 24 卷

页码：

收录情况： ESCI（2025版）

摘要： Precise prediction of rice seed viability is fundamental to enhancing grain yield and ensuring food quality. Traditional viability detection methods are time-consuming, labor-intensive, and seed-consuming, necessitating the development of non-destructive, intelligent, high-throughput detection technologies. Hyperspectral imaging (HSI), as a non-destructive technique, has been widely applied to seed viability detection. The spectral information from different seed components, as well as the number of varieties/germplasms, are key factors influencing detection performance. Therefore, we propose a Mobile Inverted Bottleneck U-Net (MBUNet) for segmenting the combined embryo and endosperm regions in rice hyperspectral images, enabling analysis of how spectral information from different seed components affects model performance. Furthermore, we propose a novel Dual Branch Spectral Transformer Network (DBST-Net) to effectively capture spectral differences between viable and non-viable seeds across 600 naturally aged rice varieties/germplasms, encompassing 58,129 samples. The DBST-Net integrates a multiscale CNN module and a Transformer module to extract local and global spectral features, respectively, while the Multi-Level Feature Fusion Attention (MFFA) module enables deep interaction between these features. Experimental results show that MBUNet achieves a mean Intersection over Union (mIoU) of 95.62 % on the rice seed hyperspectral dataset, outperforming five other segmentation models. Under imbalanced condition of viable and non-viable samples, the DBST-Net achieves the highest classification accuracy of 97.54 % among eight classification methods. The viability-sensitive region analysis reveals that the spectral information of the embryo exhibits the strongest correlation with viability prediction. The transfer experiment demonstrates that the DBST-Net possesses excellent generalization capability. This study provides a novel solution for non-destructive intelligent viability detection of large-sample cross-variety/germplasm rice seeds, offering technical support for safeguarding grain quality.

分类号：

  • 相关文献

[1]Spectral Kolmogorov-Arnold Transformer for few-shot rice germplasm viability detection using hyperspectral imaging. Hubo Xu,Ziqiang Wang,Fei Li,Wei Zhang,Han Zhang,Hailiang Zhang,Weiqi Li,Zhaohui Xue,Xia Xin. 2025

[2]Enhanced detection of Aspergillus flavus in peanut kernels using a multi-scale attention transformer (MSAT): Advancements in food safety and contamination analysis. Zhen Guo,Jing Zhang,Haifang Wang,Haowei Dong,Shiling Li,Xijun Shao,Jingcheng Huang,Xiang Yin,Qi Zhang,Yemin Guo,Xia Sun,Ibrahim Darwish. 2024

[3]A novel dual-branch spatial-spectral attention fusion model and method: A case study for the detection of nicotine content in tobacco leaves. Fukang Xing,Rongguang Zhu,Shichang Wang,Lingfeng Meng,Fujia Dong,Songfeng Wang,Jie Ren,Zongxiu Bai,Yapeng Kang. 2025

[4]Activity levels and expression of antioxidant enzymes in the ascorbate-glutathione cycle in artificially aged rice seed. Yin, Guangkun,Xin, Xia,Song, Chao,Chen, Xiaoling,Zhang, Jinmei,Wu, Shuhua,Li, Ruifang,Liu, Xu,Lu, Xinxiong. 2014

[5]番石榴实蝇性别决定基因Bcotra和Bcotra-2的克隆、序列特征及表达分析. 武强,和丹阳,张桂芬,李建伟,万方浩. 2015

[6]基于卷积神经网络与注意力机制的高光谱图像分类. 高玉鹏,闫伟红,潘新. 2024

[7]融合Transformer和LSTM的蓝莓根区土壤含水量预测模型. 王亿,曹姗姗,孙伟,胡博,古丽米拉·克孜尔别克,孔繁涛. 2024

[8]基于RT-WEDT的麦穗检测与计数方法. 李婕,杨子豪,郑权,乔江伟,涂静敏. 2024

[9]FE-TCM: Filter-Enhanced Transformer Click Model for Web Search. Yingfei Wang,Jianping Liu,Jian Wang,Xiaofeng Wang,Meng Wang,Xintao Chu. 2023

[10]A Topicality Relevance-Aware Intent Model for Web Search. Wang, Meng,Liu, Jianping,Wang, Jian,Wang, Yingfei,Chu, Xintao. 2023

[11]Protein-protein interaction and site prediction using transfer learning. Tuoyu Liu,Han Gao,Xiaopu Ren,Guoshun Xu,Bo Liu,Ningfeng Wu,Huiying Luo,Yuan Wang,Tao Tu,Bin Yao,Feifei Guan,Yue Teng,Huoqing Huang,Jian Tian. 2023

[12]TransVAE-DTA: Transformer and variational autoencoder network for drug-target binding affinity prediction. Zhou C.,Li Z.,Song J.,Xiang W.. 2024

[13]AECA-FBMamba: A Framework with Adaptive Environment Channel Alignment and Mamba Bridging Semantics and Details. Xin Chai,Wenrong Zhang,Zhaoxin Li,Ning Zhang,Xiujuan Chai. 2025

[14]DeepAT: A Deep Learning Wheat Phenotype Prediction Model Based on Genotype Data. Jiale Li,Zikang He,Guomin Zhou,Shen Yan,Jianhua Zhang. 2024

[15]Neuro-sensory evaluation of citrus flavors: A hierarchical spatial fusion approach for emotion-driven food innovation. Zhao, Qian,Yang, Peilin,Liang, Yushen,Xu, Zhenzhen,Chen, Jianle,Chen, Shiguo,Ye, Xingqian,Cheng, Huan. 2025

[16]A Self-Supervised Pre-Trained Transformer Model for Accurate Genomic Prediction of Swine Phenotypes. Weixi Xiang,Zhaoxin Li,Qixin Sun,Xiujuan Chai,Tan Sun. 2025

[17]Transformer is involved in female sex determination in Bemisia tabaci. Liu, Yating,Wang, Wenlu,Wang, Yina,Zhou, Xuguo,Xie, Wen,Zhang, Youjun. 2025

[18]Optimization Of Fish Quality By E.valuation Of Total Volatile B asic Nitrogen (Tvb-N) And Texture Profile Analysis (Tpa) By Near-Infrared (Nir) Hyperspectral Imaging. Zhao, Junbo,Zhang, Yituo,Wang, Xue,Shan, Jiajia,Han, Shuqing. 2019

[19]Identification Of Adulterated Cooked Millet F.lour With Hyperspectral Imaging A nalysis. Xuan, Guantao,Shao, Yuanyuan,Shao, Yuanyuan,Hu, Zhichao,Gao, Xuemei. 2018

[20]Early Detection Of Aspergillus Parasiticus I.nfection In Maize Kernels U sing Near-Infrared Hyperspectral Imaging And Multivariate Data Analysis. Wang, Wei,Zhao, Xin,Li, Chunyang,Chu, Xuan,Kimuli, Daniel. 2017

作者其他论文 更多>>