Project highlights

  • Trees and plants are vital for biodiversity and human well-being. AI technology can help monitor and protect them from the spread of diseases, ensuring long-term environmental and societal benefits. 
  • Developing AI-driven tools, including computer vision and machine learning algorithms, will enable early detection and prediction of pest and disease spread, provide quantitative assessments, and support rapid, cost-effective interventions.  
  • Collaborating with Plant Health at Defra and stakeholders, insights gained can drive informed decision-making, ensuring that the benefits of proactive actions outweigh the associated costs and efforts. 

Overview

Plants and trees are vital to the UK, providing £15.7 billion annually in economic, environmental, and social benefits, and are key to achieving Net Zero by 2050. However, global plant trade and climate change threaten sectors like agriculture and forestry. The UK Plant Health Risk Register has identified over 700 pests and pathogens as biosecurity risks, as shown in Fig1. Since 1976, pests and tree diseases have cost the UK £5-13 billion. Ash dieback alone could kill 80% of UK ash trees and cost £15 billion, severely impacting this iconic species.  

In partnership with Plant Health at Defra, this project aims to develop AI, computer vision, and deep learning technology to detect plant and tree health issues (e.g. pests and diseases), aligning with the UK Plant Biosecurity Strategy. The solution will protect Great Britain’s plants, support effective risk management, optimize care practices, and enable safe trade. 

To achieve cost-effective plant and tree health monitoring and disease detection, the project will leverage digital imaging technology combined with advanced AI analytics. Cameras, including drone-mounted RGB, multispectral, and hyperspectral sensing data, will be used. Multispectral and hyperspectral imaging can detect subtle plant health changes, such as stress or nutrient deficiencies, that are invisible to the naked eye. Drones and satellite images will offer large-scale aerial monitoring, providing early warnings of issues, hotspots, or disease outbreaks. 

Deep learning has revolutionized digital image processing, surpassing traditional machine learning methods. By leveraging deep neural networks, it can learn intricate features directly from vast amounts of multimodal image and spectral data collected from ground-based systems, satellites, and drones.  

Developed deep learning and AI algorithms will analyze images to recognize patterns and symptoms such as discoloration, wilting, or spotting. These algorithms will classify plant health by species and disease and assess damage severity. By leveraging advanced AI and deep learning technology, the project will enable more accurate and automated plant health analysis, promoting sustainable practices. 

Circular diagram listing in a clockwise direction forestry, field grown crop, ornamental crop, greenhouse crop, and orchard crop and the number of pests that threaten or endanger each of them.

Fig1. Potential pest threats screened through the UK Plant Health Risk Register https://www.gov.uk/government/publications/plant-biosecurity-strategy-for-great-britain-2023-to-2028/plant-biosecurity-strategy-for-great-britain-2023-to-2028 

Host

Loughborough University

Theme

  • Climate and Environmental Sustainability
  • Organisms and Ecosystems

Supervisors

Project investigator

Co-investigators

How to apply

Methodology

Datasets and preprocessing: public datasets and data from Plant Health at Defra and collaborators will be used, including high-resolution plant images and multispectral/hyperspectral satellite data. Techniques like image augmentation, normalization, and spectral band extraction will enhance data quality for AI learning. 

AI and deep learning model development: compact CNNs and segmentation models will extract features, detect regions, and classify diseases. Attention mechanisms will focus on key areas, while transfer learning and semi-supervised learning will adapt models for new diseases with limited data. GANs will generate synthetic data when real-world samples are scarce.  

Model evaluation: AI models will be validated using various performance metrics and rigorously tested with real-world data and case studies from Defra to ensure practical applicability.
 

Domain knowledge: Experts (Defra) validate data, interpret results, identify biases, and guide feature importance based on plant biology and disease knowledge. This ensures models are accurate, reliable, and practical for real-world use.  

Training and skills

DRs will be awarded CENTA Training Credits (CTCs) for participation in CENTA-provided and ‘free choice’ external training. One CTC can be earned per 3 hours training, and DRs must accrue 100 CTCs across the three and a half years of their PhD.  

We will provide the following training:  

  • Deep learning including essential concepts and models like CNNs, GANs, transfer learning. 
  • Programming using Python, TensorFlow, and PyTorch. 
  • Statistical analysis skills, data processing and evaluation metrics. 
  • Understanding research methodology, experiment design, project management, and ethics in AI. 
  • The development of communication and collaboration skills, including presenting research, writing papers, and engaging with stakeholders.  

The Loughborough University offers a range of AI modules, high-spec GPU A100 computers, along with research skill and career training. Plant Health at Defra provides expertise and supervision in data understanding, validation, and statistical analysis, along with training in plant/tree health, disease diversity, biosecurity, and sustainable practices.   

Partners and collaboration

Prof. Baihua Li, an expert in AI at Loughborough University with 20+ years of experience in computer vision and machine learning, has published 120+ papers and supervised 20+ PhDs. She leads AI projects in agriculture, including a £2.5M UKRI/EPSRC-funded project on AI and digital twins, and an Innovate UK project on AI herbicide inspectors.  

Sam Grant, Plant Health Statistician at Defra, leads a team providing statistics and analysis for better pest and disease management.  Through collaboration with stakeholders and AI, we aim to enhance early tree disease detection, provide insights, and enable faster, cost-effective monitoring, streamlining the current analytics process. 

Further details

For further information about this project, please contact Professor Baihua Li ([email protected] ) at Loughborough University

To apply to this project: 

  • You must include a CV with the names of at least two referees (preferably three) who can comment on your academic abilities.  
  • Please submit your application and complete the host institution application process via: https://www.lboro.ac.uk/study/postgraduate/apply/research-applications/   The CENTA Studentship Application Form 2025 and CV, along with other supporting documents required by Loughborough University, can be uploaded at Section 10 “Supporting Documents” of the online portal.  Under Section 4 “Programme Selection” the proposed study centre is Central England NERC Training Alliance.  Please quote CENTA 2025-LU7 when completing the application form. 
  • For further enquiries about the application process, please contact the School of Social Sciences & Humanities ([email protected]). 

Applications must be submitted by 23:59 GMT on Wednesday 8th January 2025.  

Possible timeline

Year 1

  • Literature review on plant and tree health; deep learning-based detection, classification and segmentation; define research objectives. 
  • Data collection and preprocessing (e.g. RGB, satellite, ground, and drone data); training in relevant AI and statistical methods. 
  • Development of preliminary AI models (e.g., CNNs for image analysis); initial experiments with data. 
  • Refine models based on initial results; attend relevant conferences or workshops. 

Year 2

  • Expand model development, integrating multimodule data for complex disease detection. 
  • Conduct validation and optimization of AI models; collaborate with domain experts (Defra). 
  • Implement transfer learning and semi-supervised learning for enhanced performance. 
  • Publish initial findings; present at conferences; refine models based on feedback. 

Year 3

  • Explore novel data and improve deep learning models (e.g. attention mechanism);  
  • Case studies and real-world testing of models.  
  • Evaluate model generalization to different disease types, species and environments.  
  • Complete final optimizations; integrate GANs for synthetic data generation if needed. 
  • Finalize thesis writing; submit publications; prepare for PhD viva. 

 

Year 4 (optional): Further refinement of research, publications and viva. 

Further reading

Gupta, Akshit, et al. (2024). GreenScan: Towards large-scale terrestrial monitoring the health of urban trees using mobile sensing. IEEE Sensors Journal, vol. 24, no. 13, pp. 21286-21299. https://ieeexplore.ieee.org/document/10529969/. 

Plant health research and development plan (Accessed 2024). https://www.gov.uk/government/publications/plant-health-research-and-development-plan-2023-to-2028/plant-health-research-and-development-plan 

Policy paper: Plant biosecurity strategy for Great Britain (2023 to 2028) (accessed 2024), https://www.gov.uk/government/publications/plant-biosecurity-strategy-for-great-britain-2023-to-2028/plant-biosecurity-strategy-for-great-britain-2023-to-2028 

Tito Arevalo-Ramirez, Anali Alfaro, José Figueroa, Mauricio Ponce-Donoso, Jose M. Saavedra, Matías Recabarren, José Delpiano (2024). Challenges for computer vision as a tool for screening urban trees through street-view images. Urban Forestry & Urban Greening, Volume 95, 128316, https://doi.org/10.1016/j.ufug.2024.128316. 

Zhang, M and Meng, Q (2011). Automatic citrus canker detection from leaf images captured in field, Pattern Recognition Letters, 32(15), pp.2036-2046, DOI: 10.1016/j.patrec.2011.08.003.