Estimation of Botanical Composition in Mixed Clover-Grass Fields Using Machine Learning-Based Image Analysis.

Sashuang Sun,Lin Zhao,Letan Luo,Ning Liang,David Parsons,Jiang Shi,Zhao Wang,Zhiyu Zuo,Julien Morel,Zhenjiang Zhou,Yong He,Hui Fang

doi:10.3389/fpls.2021.622429

Abstract

This study aims to provide an effective image analysis method for clover detection and botanical composition (BC) estimation in clover–grass mixture fields. Three transfer learning methods, namely, fine-tuned DeepLab V3+, SegNet, and fully convolutional network-8s (FCN-8s), were utilized to detect clover fractions (on an area basis). The detected clover fraction (CFdetected), together with auxiliary variables, viz., measured clover height (Hclover) and grass height (Hgrass), were used to build multiple linear regression (MLR) and back propagation neural network (BPNN) models for BC estimation. A total of 347 clover–grass images were used to build the estimation model on clover fraction and BC. Of the 347 samples, 226 images were augmented to 904 images for training, 25 were selected for validation, and the remaining 96 samples were used as an independent dataset for testing. Testing results showed that the intersection-over-union (IoU) values based on the DeepLab V3+, SegNet, and FCN-8s were 0.73, 0.57, and 0.60, respectively. The root mean square error (RMSE) values for the three transfer learning methods were 8.5, 10.6, and 10.0%. Subsequently, models based on BPNN and MLR were built to estimate BC, by using either CFdetected only or CFdetected, grass height, and clover height all together. Results showed that BPNN was generally superior to MLR in terms of estimating BC. The BPNN model only using CFdetected had a RMSE of 8.7%. In contrast, the BPNN model using all three variables (CFdetected, Hclover, and Hgrass) as inputs had an RMSE of 6.6%, implying that DeepLab V3+ together with BPNN can provide good estimation of BC and can offer a promising method for improving forage management.

Highlights

Forage crops are the main source of nutrition for ruminant animals such as cows
Detection Performance of Clover Fraction Based on DeepLab V3+, SegNet, and Fully Convolutional Network-8s Models
With a high clover content in the sample images, the detected clover fractions (CFs) were closer to the true CFs by using the fully convolutional network-8s (FCN-8s) network (Figures 6E2,E3)

Summary

Introduction

High-quality forages promote the growth of ruminants and result in more efficient production and high-quality animal products. For either grazing or harvest, include a mixture of grass and clover, or other legumes (Steinshamn and Thuen, 2008). A grass–legume polyculture can use Forage Botanical Composition Estimation the resources of water, soil nutrients, space, light, and heat more efficiently and can improve the yield and quality of the forage. A forage with a high botanical composition (BC) can provide a high quality of feed for livestock. Accurate estimation of BC (i.e., the fraction of clover by dry weight, hereinafter referred to as BC) in the mixed clover–grass fields is necessary for fertilization decision making (Nyfeler et al, 2011), estimation of forage quality (Parsons et al, 2013), and general assessment of the performance of grassland

Objectives

Methods

Results

Discussion

Conclusion