A transformer-based approach for early prediction of soybean yield using time-series images
Crop yield prediction which provides critical information for management decision-making is of significant importance in precision agriculture. Traditional manual inspection and calculation are often laborious and time-consuming. For yield prediction using high-resolution images, existing methods, e...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2023-06-01
|
| Series: | Frontiers in Plant Science |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fpls.2023.1173036/full |
| _version_ | 1827920962263187456 |
|---|---|
| author | Luning Bi Owen Wally Guiping Hu Albert U. Tenuta Yuba R. Kandel Daren S. Mueller |
| author_facet | Luning Bi Owen Wally Guiping Hu Albert U. Tenuta Yuba R. Kandel Daren S. Mueller |
| author_sort | Luning Bi |
| collection | DOAJ |
| description | Crop yield prediction which provides critical information for management decision-making is of significant importance in precision agriculture. Traditional manual inspection and calculation are often laborious and time-consuming. For yield prediction using high-resolution images, existing methods, e.g., convolutional neural network, are challenging to model long range multi-level dependencies across image regions. This paper proposes a transformer-based approach for yield prediction using early-stage images and seed information. First, each original image is segmented into plant and soil categories. Two vision transformer (ViT) modules are designed to extract features from each category. Then a transformer module is established to deal with the time-series features. Finally, the image features and seed features are combined to estimate the yield. A case study has been conducted using a dataset that was collected during the 2020 soybean-growing seasons in Canadian fields. Compared with other baseline models, the proposed method can reduce the prediction error by more than 40%. The impact of seed information on predictions is studied both between models and within a single model. The results show that the influence of seed information varies among different plots but it is particularly important for the prediction of low yields. |
| first_indexed | 2024-03-13T04:19:10Z |
| format | Article |
| id | doaj.art-d5aceb4ff2bd4158b9f0ebbb6814cf36 |
| institution | Directory Open Access Journal |
| issn | 1664-462X |
| language | English |
| last_indexed | 2024-03-13T04:19:10Z |
| publishDate | 2023-06-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Plant Science |
| spelling | doaj.art-d5aceb4ff2bd4158b9f0ebbb6814cf362023-06-20T15:14:55ZengFrontiers Media S.A.Frontiers in Plant Science1664-462X2023-06-011410.3389/fpls.2023.11730361173036A transformer-based approach for early prediction of soybean yield using time-series imagesLuning Bi0Owen Wally1Guiping Hu2Albert U. Tenuta3Yuba R. Kandel4Daren S. Mueller5Department of Industrial and Manufacturing Systems Engineering, Iowa State University, Ames, IA, United StatesAgriculture and Agri-Food Canada, Harrow Research and Development Centre, Harrow, ON, CanadaDepartment of Industrial and Manufacturing Systems Engineering, Iowa State University, Ames, IA, United StatesOntario Ministry of Agriculture, Food and Rural Affairs, Ridgetown, ON, CanadaDepartment of Plant Pathology and Microbiology, Iowa State University, Ames, IA, United StatesDepartment of Plant Pathology and Microbiology, Iowa State University, Ames, IA, United StatesCrop yield prediction which provides critical information for management decision-making is of significant importance in precision agriculture. Traditional manual inspection and calculation are often laborious and time-consuming. For yield prediction using high-resolution images, existing methods, e.g., convolutional neural network, are challenging to model long range multi-level dependencies across image regions. This paper proposes a transformer-based approach for yield prediction using early-stage images and seed information. First, each original image is segmented into plant and soil categories. Two vision transformer (ViT) modules are designed to extract features from each category. Then a transformer module is established to deal with the time-series features. Finally, the image features and seed features are combined to estimate the yield. A case study has been conducted using a dataset that was collected during the 2020 soybean-growing seasons in Canadian fields. Compared with other baseline models, the proposed method can reduce the prediction error by more than 40%. The impact of seed information on predictions is studied both between models and within a single model. The results show that the influence of seed information varies among different plots but it is particularly important for the prediction of low yields.https://www.frontiersin.org/articles/10.3389/fpls.2023.1173036/fulltransformerimage recognitiontime-series predictionsoybean yield predictiondeep learning |
| spellingShingle | Luning Bi Owen Wally Guiping Hu Albert U. Tenuta Yuba R. Kandel Daren S. Mueller A transformer-based approach for early prediction of soybean yield using time-series images Frontiers in Plant Science transformer image recognition time-series prediction soybean yield prediction deep learning |
| title | A transformer-based approach for early prediction of soybean yield using time-series images |
| title_full | A transformer-based approach for early prediction of soybean yield using time-series images |
| title_fullStr | A transformer-based approach for early prediction of soybean yield using time-series images |
| title_full_unstemmed | A transformer-based approach for early prediction of soybean yield using time-series images |
| title_short | A transformer-based approach for early prediction of soybean yield using time-series images |
| title_sort | transformer based approach for early prediction of soybean yield using time series images |
| topic | transformer image recognition time-series prediction soybean yield prediction deep learning |
| url | https://www.frontiersin.org/articles/10.3389/fpls.2023.1173036/full |
| work_keys_str_mv | AT luningbi atransformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT owenwally atransformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT guipinghu atransformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT albertutenuta atransformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT yubarkandel atransformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT darensmueller atransformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT luningbi transformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT owenwally transformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT guipinghu transformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT albertutenuta transformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT yubarkandel transformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages AT darensmueller transformerbasedapproachforearlypredictionofsoybeanyieldusingtimeseriesimages |