Applications of AI in Plant Phenotyping
Steering committee
Associate Prof. Dr. Arti Singh, Iowa State University, USA (Chair)
Germano Gomes, VIB Agro-Incubator, BEL (co-Chair)
Summary of the Working Group
To maintain progress in agriculture, specifically in achieving climate resilience and sustainability in crop production, it is essential that we deploy emerging computer vision (CV) and more broadly Artificial Intelligence (AI) models. These fields are rapidly advancing thanks to powerful graphics processing units (GPUs), very large datasets, and open-source libraries. Deep learning and the Transformer architecture used in large language models (LLMs) have garnered significant attention, with applications spanning multiple industries, such as automotive and medical sciences. In agriculture and plant sciences, we have witnessed the transition from supervised learning to self-supervised learning techniques with transfer learning, laying the foundation for the successful use of these domain specific models. The Applications of AI in Plant Phenotyping working group will coordinate efforts to address real-world, use-defined challenges that enable researchers and empower farmers. There are numerous opportunities to leverage CV and more broadly AI models to directly impact agriculture:
- Genotype-environment-phenotype association - Understanding the relationship between plant genotype plus environment and phenotype is crucial for plant breeding and crop improvement. AI models can be used to predict plant phenotypes based on genetic and environmental data, potentially helping in identifying desirable traits for breeding and crop improvement.
- Yield prediction - Accurately forecasting crop yield is essential in agriculture. ML models can be trained to predict yield based on factors like weather conditions, soil properties, and plant traits.
- Plant stress detection and quantification (biotic factors) - Diseases, insect pests, and weeds can significantly reduce crop yield and quality.
- Plant stress detection and quantification (abiotic factors) - Environmental stresses like nutrient deficiency, herbicide injury, freeze injury, flooding, drought, salinity, and extreme temperatures can impact plant growth and yield.
- Plant growth monitoring - Assessing morphological and physiological traits is a vital part of plant phenotyping. For tasks 3, 4, and 5, satellite, drone, or ground rover-based imaging with sensors like thermal, 3D, NIR, and hyperspectral can be combined with CV and more broadly AI to study these traits and help farmers take preventive measures to mitigate abiotic and biotic stresses and monitor plant growth and development.
News
[27.02.2024] Official Working Group kick-off: Applications of Artificiel Intelligence in Plant Phenotyping