Case Study

From Field Photos to Precision Farming Intelligence at Scale

Taranis, a global leader in digital agronomy, harnesses AI to provide real-time crop insights.. That's where IndiVillage came in. Operating in regions like the USA, Brazil, and Australia, it offers multi-layered image analysis from satellite to leaf-level. Taranis partnered with IndiVillage to deliver on a mutual promise of proactive crop intelligence that prevents problems before they impact yield.

Taranis

Taranis, a global leader in digital agronomy, harnesses AI to provide real-time crop insights. That's where IndiVillage came in. Operating in regions like the USA, Brazil, and Australia, it offers multi-layered image analysis from satellite to leaf-level. Taranis partnered with IndiVillage to deliver on a mutual promise of proactive crop intelligence that prevents problems before they impact yield.

460+

weed species tagged

4.5M+

images annotated to date

50%

annual increase in annotation capacity

Challenge

Massive Scale, Minute Detail

Taranis needed to process millions of high-resolution aerial images to detect early signs of weeds, pests, diseases, and nutrient issues. But each image brought variation of different crops, growth stages, lighting, and regional species. Tagging anomalies required agricultural knowledge and pixel-level accuracy. With over 460 weed species alone, image ambiguity and seasonal variance made standardization tough.

Taranis also required a team that could quickly adapt to fluctuating image volumes, handle edge-case data with context, and deliver high-throughput tagging across diverse geographies. Consistency, agility, and precision were non-negotiable for success. To ensure consistency, the solution needed to combine subject-matter expertise, multi-level QC, and rapid scalability.

Solution

Expert Tagging That Trains Smarter AI

IndiVillage deployed a 70-member expert team with domain expertise across 30 crops. A dual-mode workflow was built on Dataloop and Jira, AI2 tagging marked visible anomalies for customer reports, while AI tagging captured pixel-perfect details to train detection models. Tags captured weeds, pests, and nutrient deficiencies with high specificity. Weeds were categorized by type and shape; diseases and insect damage were tagged based on visual crop stress cues.

All image tasks were managed through Dataloop and Jira, ensuring smooth transitions from upload to QC and final delivery. To ensure reliability, we implemented a 3-level QC framework, L1 tagging, L2 internal review, and L3 final QC via Label Filter View. Daily sprints and task pipelines ensured images moved from upload to delivery smoothly, even during volume spikes. Regular alignment with Taranis helped fine-tune taxonomy, adjust tagging strategies, and prioritize accuracy over speed.

Results

Scaling Insight, Strengthening Yields Worldwide

With IndiVillage's precision-driven tagging, Taranis dramatically scaled its AI training data. Over 4.5M+ images were annotated to date, with annotation capacity growing 50% annually.

The high-quality annotations enabled faster and more accurate anomaly detection across nine global regions, supporting proactive farm health decisions and boosting customer confidence.

The partnership transformed data into action. By combining detailed ground truth with scalable human expertise, IndiVillage enabled Taranis to operationalize intelligence at a global level, bridging AI performance with real-world agricultural impact.

“IndiVillage is doing their own internal quality assurance efforts, building dashboards and new tools - they are the only one of our suppliers to do this proactively. Our markets have different characters, geographic, different weeds, different agriculture practices... and IndiVillage's outcome is superior to the rest.”
Idan HararyDirector of Ag, Taranis