Global tech giant Google is building a digital agri-stack using satellite imagery and has started with a base layer that will help identify farm boundaries, a senior official said on Saturday.

Data and analysis of this base layer of the stack can be used in a slew of applications like bettering the subsidy payments, farm insurance or farm loans, Manish Gupta, a senior director at Google DeepMind said.

“We have built the first such model that uses satellite imagery analysis and can now start to identify field boundaries based on usage pattern, start to identify what crops are being grown and so on,” Gupta said, speaking at the Mumbai Tech Week here.

The model has taken inspiration from UIDAI’s Aadhar, ascribing a unique ID to each farm and is intended to act as a base layer of the “digital agri stack”, he said.

Gupta said it is upon the start-ups and other agri-tech players to build solutions which will provide loan crop insurance and also subsidies using the data churned out by the platform.

Pointing out the relevance of such a platform, he said 40 per cent of the Indian population are agriculturists and the annual farm lending alone is a $550 billion opportunity with a lot of it cornered by the informal sector lenders.

The comments come at a time when the Indian banking system is trying to improve its agri-lending system. Largest lender SBI’s chairman C S Setty spoke this week about the bank mulling to use satellite imagery-based inputs for making its agri lending processes more efficient.

• Also read: Technicolor India shuts operations amid global troubles for the company; netizens blame AI

Meanwhile, Gupta also rued that the amount of content available in Indian languages is very limited and Google is trying to address the same with its initiatives.

He said about 10 per cent of the overall 8 billion people speak Hindi, but the language accounts for only 0.1 per cent of the online content.

It has launched the “Vaani” project aimed at bridging this gap under which aspires to collect speech data in form of voice recordings from across over 700 districts.

The company has launched the first phase of the project under which it has collected 14,000 hours of audio data from 80 districts with people speaking in 59 languages, he said.

He also said that Indians underappreciate the relevance of artificial intelligence to expand the boundaries of science, and pointed to the work by DeepMind which led to a Nobel Prize in Chemistry.