您好,欢迎访问北京市农林科学院 机构知识库!

Maize Anthesis-Silking Interval Estimation via Image Detection under Field Rail-Based Phenotyping Platform

文献类型: 外文期刊

作者: Zhuang, Lvhan 1 ; Wang, Chuanyu 1 ; Hao, Haoyuan 1 ; Song, Wei 5 ; Guo, Xinyu 1 ;

作者机构: 1.Beijing Acad Agr & Forestry Sci, Informat Technol Res Ctr, Beijing 100097, Peoples R China

2.Natl Engn Res Ctr Informat Technol Agr, Beijing Key Lab Digital Plant, Beijing 100097, Peoples R China

3.Hainan Univ, Sch Mech & Elect Engn, Haikou 570228, Peoples R China

4.Beijing Technol & Business Univ, Sch Comp & Artificial Intelligence, Beijing 100048, Peoples R China

5.Hebei Acad Agr & Forestry Sci, Inst Cereal & Oil Crops, Key Lab Crop Genet & Breeding Hebei Prov, Shijiazhuang 050031, Peoples R China

关键词: field rail-based phenotyping platform; anthesis-silking interval (ASI); image detection; YOLOv8

期刊名称:AGRONOMY-BASEL ( 影响因子:3.4; 五年影响因子:3.8 )

ISSN:

年卷期: 2024 年 14 卷 8 期

页码:

收录情况: SCI

摘要: The Anthesis-Silking Interval (ASI) is a crucial indicator of the synchrony of reproductive development in maize, reflecting its sensitivity to adverse environmental conditions such as heat stress and drought. This paper presents an automated method for detecting the maize ASI index using a field high-throughput phenotyping platform. Initially, high temporal-resolution visible-light image sequences of maize plants from the tasseling to silking stage are collected using a field rail-based phenotyping platform. Then, the training results of different sizes of YOLOv8 models on this dataset are compared to select the most suitable base model for the task of detecting maize tassels and ear silks. The chosen model is enhanced by incorporating the SENetv2 and the dual-layer routing attention mechanism BiFormer, named SEBi-YOLOv8. The SEBi-YOLOv8 model, with these combined modules, shows improvements of 2.3% and 8.2% in mAP over the original model, reaching 0.989 and 0.886, respectively. Finally, SEBi-YOLOv8 is used for the dynamic detection of maize tassels and ear silks in maize populations. The experimental results demonstrate the method's high detection accuracy, with a correlation coefficient (R2) of 0.987 and an RMSE of 0.316. Based on these detection results, the ASI indices of different inbred lines are calculated and compared.

  • 相关文献
作者其他论文 更多>>