Scaling Pest Detection With AI Across 300K Hectares

FMC faced a complex challenge of training AI models that could accurately detect and classify pests in dynamic agricultural environments.With millions of images captured across different regions and crop cycles, the company needed a reliable way to benchmark the performance of their models against human tagging. The objective was to streamline annotation workflows, improve tagging accuracy, and scale predictive insights without compromising quality. With millions of data points and evolving pest patterns, precise identification of pest species was critical for meaningful predictions.

IndiVillage implemented a custom AI training workflow anchored in Microsoft Custom Vision. A dedicated project was set up to manage large-scale image annotation of pest species critical to FMC’s agricultural use cases. Our trained annotation team carried out precise bounding box labeling for each pest instance and tagged insect species with high fidelity, following an expert-defined taxonomy to ensure biological and functional accuracy. Consistency was a key focus. Uniform tagging standards were enforced across the dataset to deliver structured, reliable data. By integrating manual precision with AI-assisted review tools, we not only accelerated annotation throughput but also elevated data quality. This hybrid approach empowered FMC to benchmark human vs. AI accuracy more effectively, enabling iterative model refinement and stronger, real-world pest detection performance.

Through this collaboration, FMC’s Arc™ Farm Intelligence platform has significantly scaled its reach, now covering over 300,000 hectares with the support of seven advanced AI models achieving over 96% precision in pest identification. IndiVillage played a key role in refining and validating the pest tagging workflows, ensuring high-quality data served as the backbone for model training and deployment. By embedding these AI-driven insights directly into field operations, FMC is empowering farmers to adopt more sustainable, efficient, and cost-effective approaches to pest management.