2018
Using engineering tools to identify and quantify biotic and abiotic stress in soybean for customizable agriculture production
Contributor/Checkoff:
Category:
Sustainable Production
Keywords:
Abiotic stressAgricultureLand Use Water supply
Lead Principal Investigator:
Arti Singh, Iowa State University
Co-Principal Investigators:
Project Code:
Contributing Organization (Checkoff):
Leveraged Funding (Non-Checkoff):
Results from this project helped me to get my new award(~500K) from USDA-NIFA -FACT: "A scalable Cyber ecosystem for acquisition, Curation and Analysis of multi-spectral UAV image data
Show More
Institution Funded:
Brief Project Summary:

Crop yields are inherently limited by plant stresses (biotic and abiotic). Plant breeders have protected yield from plant stress losses by incorporating resistance genes and developing more resilient cultivars. State-of-the-art High Throughput Phenotyping has unlocked new prospects for field-based phenotyping. What is currently lacking is methodology to quickly screen HTP images into easy-to-use tools that help identify, detect, classify and predict plant diseases. This research aims to use hyperspectral camera and spectro-radiometer to develop disease signatures to distinguish among SCN, SDS, BSR, charcoal rot and IDC; develop algorithms to differentiate diseases with confounding symptoms; develop predictions for disease onset using "disease signatures” and develop an algorithm to count SCN eggs under the microscope rapidly and accurately manner.

Key Benefactors:
farmers, agronomists, Extension agents, soybean breeders, seed companies

Information And Results
Project Deliverables

2016-17: Inoculate and/or rate plants or plots with SCN, SDS, BSR, and IDC (in field and/or indoors) to collect hyperspectral images. Human disease scoring will be done on a crop canopy. Data will be stored on high capacity drives for further processing and analysis. Algorithms to differentiate various diseases will be developed.
Oct 2017-December 18: Phenotyping of 300 soybean genotypes including 30 NAM parents and 270 diverse PI accessions for SCN in GH and field, as well as all previously reported SCN sources in the large genotype panel for genomic analysis.
(a) SCN screening in GH
(b) SCN screening in field
December 2018 – March 2019: SCN screening. Data preparation for association and genomic selection studies.

Final Project Results

Update:
The goals of this research is to identify and differentiate biotic stresses under soybean production using hyperspectral camera and remote sensing spectro-radiometer to pave the way for farmers and researchers to use sensors, smartphones and aerial imagery for crop management.
One of our long-term goals is to make smartphone apps for farmers to assist in scouting, enabling the farmer to use their smartphone to determine the presence, severity of specific diseases in order to make strategic decisions on disease control.
Research Progress
• Fall of 2018, Work on multiple stress detection study got published in “Proceedings of National Academy of Sciences” (PNAS) on: An Explainable Deep Machine Vision Framework for Plant Stress Phenotyping
• Work on Hyperspectral disease signature on charcoal rot got published in plant methods journal on “Hyperspectral band selection using genetic algorithm and support vector machines for early identification of charcoal rot disease in soybean.
• Multiple stress images were collected in summer of 2018 to further expand the ICQP paradigm using machine learning to automate phenotyping.
• Fall of 2018, Review of literature for SCN GWAS in soybean and project plan created to analyze in germplasm greenhouse. Experiment will start in green house in spring 2018.
• Fall of 2016, images of charcoal rot resistant and susceptible plants were taken and hyperspectral and spectoradiometer data was collected. The results of charcoal rot hyperspectral signatures were presented in 4th International Plant Phenotyping Network Symposium El Batan, Mexico
• In 2016, more than 25,000 leaflet images were collected from Iowa (USA) fields for five biotic stresses (bacterial leaf blight, bacterial pustule, frogeye leaf spot, Septoria brown spot and SDS) and three abiotic stresses (IDC, potassium deficiency and herbicide injury) as well as healthy leaflets on soybean using a standard imaging protocol. The Deep Convolution Neural Network (DCNN) was designed to automatically differentiate images of eight different stresses. The results of this work were presented in 4th International Plant Phenotyping Network Symposium El Batan, Mexico
• In 2016, a smartphone app (for automated phenotyping) was demonstrated to IA farmers during the ISA board members tour to Iowa State University.
• Spring 2017, hyperspectral disease signatures of root rot disease under progress.
Papers:
1. Ghoshal S, D Blystone, AK Singh, B Ganapathysubramanian, A Singh*, S Sarkar*. 2018. Bringing consistency to plant stress phenotyping through an explainable deep machine vision framework. Proceedings of the National Academy of Sciences. 115 (18) 4613-4618; DOI: 10.1073/pnas.1716999115
2. Singh AK, B Ganapathysubramanian, S Sarkar*, A Singh*. 2018. Deep learning for plant stress phenotyping: trends and future perspectives. Trends in Plant Science 23(10): 883-898 https://doi.org/10.1016/j.tplants.2018.07.004
3. Nagasubramanian K, S Jones, S Sarkar, AK Singh, A Singh*, B Ganapathysubramanian*. 2018. Hyperspectral band selection using genetic algorithm and support vector machines for early identification of charcoal rot disease in soybean stems. Plant Methods 14:86 https://doi.org/10.1186/s13007-018-0349-9
4. Akintayo A, GL Tylka, AK Singh, B Ganapathysubramanian, A Singh*, S Sarkar*. 2018. A deep learning framework to discern and count microscopic nematode eggs. Scientific Reports. 8: 9145 (Available online: 10.1038/s41598-018-27272-w).
5. Homagni Saha, Tianshuang Gao, Hamid Emadi, Zhanhong Jiang, Arti Singh, Baskar Ganapathysubramanian, Soumik Sarkar, Asheesh Singh, Sourabh Bhattacharya. 2017. Autonomous Mobile Sensing Platform for Spatio-Temporal Plant Phenotyping
ASME 2017 Dynamic Systems and Control Conference. V002T21A005-V002T21A005
American Society of Mechanical Engineers
6. A real-time phenotyping framework using machine learning for plant stress severity rating in soybean. 2017. HS Naik, J Zhang, A Lofquist, T Assefa, S Sarkar, D Ackerman, A Singh, Asheesh K Singh, Baskar Ganapathysubramanian. Plant Methods. doi: 10.1186/s13007-017-0173-7
7. Computer vision and machine learning for robust phenotyping in genome-wide studies. 2017. J Zhang, HS Naik, T Assefa, S Sarkar, RVC Reddy, A Singh, B Ganapathysubramanian, AK Singh. Scientific Reports-Nature 7. doi:10.1038/srep44048
8. Machine Learning for High-Throughput Stress Phenotyping in Plants. 2016. Singh, Arti, B Ganapathysubramanium, AK Singh and S Sarkar. Trends in Plant Science 21(2): 110-124.
9. An end-to-end convolutional selective autoencoder approach to Soybean Cyst Nematode eggs detection. 2016. A Akintayo, N Lee, V Chawla, M Mullaney, C Marett, AK Singh, Arti Singh, G Tylka, B Ganapathysubramaniam, S Sarkar. . arXiv:1603.07834.

Posters presented:
1. David Blystone, Sambuddha Ghosal, Homagni Saha, Daren Mueller, Baskar Ganapathysubramanian1, Asheesh K. Singh, Arti Singh, Soumik Sarkar. High-Throughput Identification and Quantification of Multiple Stresses in Soybean using a DCNN Framework Phenome 2018, Tuscon, Arizona
2. Zhang J., Naik H.S., Assefa T., Sarkar S., Reddy R.V., Singh A., Ganapathysubramanian B. and Singh A.K. Computer vision and machine learning for robust phenotyping in genome wide association and prediction analysis in soybean. 2017. R.F. Baker Plant Breeding Symposium, Ames, IA.
3. A Akintayo, N Lee, V Chawla, MP. Mullaney, CC. Marett, A Singh, A Singh, G Tylka, B Ganapathysubramanian, S Sarkar. End-to-end convolutional selective autoencoder for Soybean Cyst Nematode eggs detection. Phenotypic prediction: image acquisition and analysis conference - February 23-25, 2016 –Iowa State University. Poster was awarded as the best graduate student poster in the conference.
4. Jones S, AK Singh, S Sarkar, B Ganapathysubramanian, D Mueller, A Singh. Hyperspectral disease signatures for detection of charcoal rot in soybean. CIMMYT 4th International Plant Phenotyping Symposium, December 13-15, 2016. Texcoco, Mexico.
5. Ghosal S, D Blystone, H Saha, D Mueller, B Ganapathysubramanian, AK Singh, A Singh, S Sarkar. An Automated Soybean Multi-Stress Detection framework using Deep Convolutional Neural Networks. CIMMYT 4th International Plant Phenotyping Symposium, December 13-15, 2016. Texcoco, Mexico.

Invited Talk
• Botany 2019 Symposia, July 27th-31st, 2019 Tucson, Arizona, “Human in the loop ML applications in high throughput plant stress phenotyping”.
• Crops 2019 organized by Hudson Alpha Institute for Biotechnology, June 3rd-6th, 2019 Huntsville, Alabama, “Artificial intelligence driven plant stress phenotyping”
• Department of Horticulture Spring Seminar Series, February 25th, 2019 Ames, Iowa, “From Mung bean to artificial intelligence”.
• Agronomy Seminar, December 6th, 2018 Ames, Iowa, “Machine Learning Driven Data Analytics for Soybean Stress Phenotyping”.
• 30th Annual Integrated Crop Management Conference, Ames, Iowa Nov 28-29, 2018. “Sensing technologies for precision plant stress phenotyping”
• IX SIGM International Symposium on Genetics and Breeding. Keynote speaker. GenMelhor at Federal University of Vicosa, Brazil, Oct 24-25, 2018. “Machine learning driven automated multi-stress soybean phenotyping” (International)
• 5th International Plant Phenotyping Symposium (IPPS). Adelaide, Australia, Oct 2-5, 2018. “Machine learning approaches for automated plant stress phenotyping” (International)
• Hermitage Research Facility, Department of Agriculture and Fisheries, Warwick, Australia, 8th October. 2018 “Mung bean breeding and AI based phenotyping” (International)
• North Central Regional Plant Introduction Station (NC7 RTAC) Meeting, Ames, Iowa, August 15-16, 2018. “Mung bean: An opportunity crop for mid-west Farmers”
• National Agriculture and Food Research Organization, Tsukuba, Japan. “Machine learning driven image-based plant stress phenotyping in plants” April 4, 2018. (International)
• 45th Asia Pacific Advanced Network Meeting on the Role of Artificial Intelligence. “High Throughput Plant Stress Identification and Quantification” 25-29 March, Singapore, 2018. https://apan.net/meetings/apan45/speaker417.html. (International)
• ISA Board Directors Event organised by Iowa Soybean Research Center (ISRC) on “ICQP Paradigm for soybean high throughput stress phenotyping, Ames, IA, January 2018
• Invited presentation/talk “Towards high throughput stress phenotyping in soybeans using machine learning” on session topic – C-08 Plant Genetic Resources- Phenotyping Plant Genetic Resources to Support Climate Smart Agriculture ASA, CSSA and SSSA, International Annul Meetings. Oct 22-25, 2017, Tampa, Florida, USA
• ISA farmer board. Soybean Phenomics : Identify problems and develop solutions integrating phenotyping, engineering and data analytics September 9, 2016


1. Established a data analytics framework for applying machine learning for soybean stress identification/detection, classification and quantification. This lays the foundation for soybean stress detection app development.
2. Resulted in novel machine learning framework for using RGB and hyperspectral imaging in soybean stress phenotyping for each stress detection.
3. Resulted in 8 high impact papers in journals like PNAS and Trends in Plant Sciences related to soybean stress phenotyping. Additionally, 10 conference proceeding papers and dozens of posters were presented at various venues.
4. Project PI gave 15 invited talks at meetings held in USA, Brazil, Australia, Japan and Singapore.

The United Soybean Research Retention policy will display final reports with the project once completed but working files will be purged after three years. And financial information after seven years. All pertinent information is in the final report or if you want more information, please contact the project lead at your state soybean organization or principal investigator listed on the project.