AgriPestForge

Controllable Generative AI Engine to Expand Automation in Agricultural Pest Detection

 

Context:

Global crop production faces losses of up to 40% annually due to pests, with invasive insects alone costing the global economy about $70 billion. Climate change exacerbates this problem by increasing the unpredictability of pest dynamics. Conventional monitoring relies on insect traps, which require manual counting and expert identification, making the process time-consuming, error-prone, and dependent on specialized knowledge. AI and Computer Vision have shown promise in automating pest detection, but extending these solutions to new crops remains costly and labor-intensive due to data bottlenecks in acquisition, representativeness, and annotation. Generative AI has emerged as a transformative field, capable of producing synthetic data, but general-purpose models fall short in specialized agricultural domains because they lack fine control and domain-specific knowledge.  

 

Goals:

The AgriPestForge project aims to advance Computer Vision Generative AI (CV-GenAI) for specialized agricultural domains by adapting general-purpose diffusion models to generate realistic, diverse, and controllable synthetic datasets for pest detection. Methodologically, the project will contribute to state-of-the-art controllable generative techniques that improve precision, control, and representativeness in synthetic data generation. Applicationally, the project will showcase these contributions through the development of a crop-agnostic Generative AI Engine that produces synthetic images of insects on sticky traps under variable real-world conditions. This engine will facilitate the training and validation of pest detection models across different crops, reducing reliance on costly and time-intensive human data collection and annotation. 

 

Impact:

By bridging methodological advances with agricultural applications, AgriPestForge will address pressing challenges in pest monitoring caused by global warming and evolving pest populations. The project’s outcomes will expand the applicability of Generative AI to real operational environments, supporting more sustainable and efficient integrated pest management strategies. Validation across key crops such as grapes, olives, and tomatoes in collaboration with industry stakeholders will reinforce the engine’s versatility and crop-agnostic nature. Ultimately, the project will strengthen multidisciplinary collaboration between AI and agronomy, contributing to more resilient food production systems while positioning Generative AI as a transformative tool in agriculture. 

 

Consult the project's Spec Sheet here:

English

 

Funded by:

For any additional information, please contact us using the inquiries form.