Agriculture is India's largest employment sector — 42% of the workforce, 16% of GDP. Artificial intelligence is being deployed across precision farming, crop disease detection, yield estimation, weather risk assessment, satellite imagery analysis, and agricultural insurance (PMFBY). Every one of these applications requires training data annotated by people who understand Indian agriculture from the field, not from textbooks.
A crop disease detection model trained on images from American or European farms will misclassify disease patterns on Indian varietals — because the disease presentations differ, the growing conditions differ, and the soil types differ. A yield estimation model trained on irrigated Punjab farms will fail in rain-fed Vidarbha. These are not model failures. They are training data failures caused by annotators who did not understand what they were annotating.
The PMFBY (Pradhan Mantri Fasal Bima Yojana) insurance scheme is deploying satellite imagery analysis and drone footage annotation at massive scale — to assess crop damage, verify land use, and process claims automatically. This creates one of the largest annotation opportunities in Indian agriculture AI, and it requires annotators with agricultural science backgrounds who can assess crop condition from imagery with the same accuracy as a field inspection.
