This Startup Built An AI-Based Grain Analyser That Will Stop Exploitation Of Helpless Farmers

We've all heard it from elders in our family. How the food they grew up eating, 50-60 years ago, was of a much higher quality than what we consume today.

One young startup is trying to restore the quality of food we eat to a higher standard, by disrupting how food grains and other agricultural produce are graded at source, not just in India but across the world, so that we can aim to eat better quality food than ever and farmers don't get a raw deal in the process.

co-founders of nebulaa

"In India there are around 12 crore farm holdings," says Tanmay Sethi from Nebulaa over the phone, "which means 12 crore crops need to be tested. More if there are different types of crops."

The problem

So the challenge during quality testing of crops, when farmers bring their crop to wholesale markets, is that you can't spend 30 minutes inspecting and analyzing every crop, claims Tanmay. Which is what the existing technology offers right now. That's where manual subjectivity creeps into the system.

Buyers just approach a farmer at the market, cursorily inspect the grain by hand and tell the price they're willing to pay to the farmer. The farmer has no way to control the transaction, they're at the mercy of traders to give them a fair price.

"It's a buyer driven ecosystem, and it's heavily biased in the favour of the buyer," Tanmay explains. These wholesale buyers or traders don't necessarily care about farmer's interests, trying to squeeze as much margin in their own favour.

The solution

This is where MATT plugs the gap and offers a radical new alternative towards food grain quality testing. Just put the sample grains in MATT and within one minute it generates the quality analysis of the product.

Nebulaa's patent-pending MATT technology utilizes machine learning and deep learning to serve results 20 times faster than a skilled human analyst. Not just speed, the results from MATT is far more reliable and repeatable, compared to any human tester ever can be.

"Because of MATT, you don't need an expert," says Tanmay, because experts don't come for free. There's a significant cost attached to human experts for food grain testing. Something as fast and accurate as MATT hasn't existed before, Tanmay claims, adding that it will revolutionize how primary machine-testing of food grain is done all over the world.

How does it work

MATT has a primary testing tray on which the sample grains are kept. Inside it are multiple cameras that take multiple images at different wavelengths. Based on this an individual grain's profile is mapped, based on an algorithm. This stage is called segmentation, which is one of the complex steps of the quality analysis process, claims Tanmay, and MATT's segmentation process is 99.9% accurate.

Thereafter an AI 3D rendering of each grain is then analyzed for defects, fungal damage, organic impurity, tearing, etc, and a mathematical model is built for each type of grain. After setting baseline defect and defining quality parameters, MATT pretty much scans and tells you the quality of any future grain sample -- of the same crop -- without the need for training. No humans involved, completely automated, Tanmay emphasizes.

Future plans for Nebulaa

After getting selected for Google's 'Solve For India' bootcamp program, Nebulaa is conducting pilot projects in districts of Telangana and also on eNAM -- a government-hosted virtual marketplace where sellers and buyers connect to buy physical agricultural produce (among other things -- to test out MATT, and they've been having a very good response, claims Tanmay.

Nebulaa is working with government and private institutions to introduce MATT for testing and grading crops at the primary market stage -- which is closest to the farm. They are also creating universal quality tests and processes, which will enable each produce to be graded scientifically like never before.

As consumers, scientifically-tested produce will empower us to make informed buying decision. Traders -- wholesalers and retailers -- won't just have to rely on gut feel to grade a crop. And most importantly, farmers can't be swindled by a biased human and rely on MATT's analysis to get credible data on the quality of grain they're producing, and how they can improve the quality of the produce in future.