Wheat Ear Recognition Based on RetinaNet and Transfer Learning.

Jingbo Li,Shuaipeng Fei,Changchun Li,Fan Ding,Jinjin Shi,Weinan Chen,Chunyan Ma,Yilin Wang,Yacong Li,Zhen Xiao

doi:10.3390/s21144845

Abstract

The number of wheat ears is an essential indicator for wheat production and yield estimation, but accurately obtaining wheat ears requires expensive manual cost and labor time. Meanwhile, the characteristics of wheat ears provide less information, and the color is consistent with the background, which can be challenging to obtain the number of wheat ears required. In this paper, the performance of Faster regions with convolutional neural networks (Faster R-CNN) and RetinaNet to predict the number of wheat ears for wheat at different growth stages under different conditions is investigated. The results show that using the Global WHEAT dataset for recognition, the RetinaNet method, and the Faster R-CNN method achieve an average accuracy of 0.82 and 0.72, with the RetinaNet method obtaining the highest recognition accuracy. Secondly, using the collected image data for recognition, the R2 of RetinaNet and Faster R-CNN after transfer learning is 0.9722 and 0.8702, respectively, indicating that the recognition accuracy of the RetinaNet method is higher on different data sets. We also tested wheat ears at both the filling and maturity stages; our proposed method has proven to be very robust (the R2 is above 90). This study provides technical support and a reference for automatic wheat ear recognition and yield estimation.

Highlights

Wheat is the largest grain crop in world trade [1,2]
In order to evaluate the performance of the method used to identify wheat ears in this paper, target recognition algorithms, regions with convolutional neural networks (R-CNN)
This paper studies the application of deep learning technology to wheat ear recognition and chooses a better recognition model for recognizing what ears are in different growth stages and the number of wheat ears in a single image

Summary

Introduction

Wheat is the largest grain crop in world trade [1,2]. With population growth and social and economic development, the demand for wheat has increased. Rapid and accurate identification and statistics of wheat ears are fundamental for crop growth monitoring and yield estimation [8]. Traditional counting methods rely on field surveys, sampling, and weighing. These methods are inefficient, costly, and difficult to determine accurate yield estimation for large areas, severely limiting their application for breeding, monitoring plant performance in crop management, or predicting grain yield. Some spike counting methods are based on wheat spike data collected at maturity and other traits, which are not suitable for early yield prediction [9]. Many studies that apply deep learning techniques [10] to unmanned aerial systems [11,12] for wheat spikelet detection under field conditions have received much attention

Objectives

Methods

Results

Discussion

Conclusion