- Team ThinkAg
Agrograde- Solving for quality issues in fresh farm produce
Agrograde uses computer vision and machine learning to analyse and grade vegetables and fruits, and a machine that sorts them by quality
“We work on post-harvest problem statements. The biggest challenge that the entire supply chain faces is trust deficit,” says 26-year-old Kshitij Thakur, Co-founder, Agrograde, a start-up that has developed solutions for quality assessment of vegetables and fruits.
The buyer does not trust the seller and vice versa, he points out. That is mainly because there is no guarantee on the quality of the produce. The vegetables and fruits may be of different sizes, grades and quality and there is none the wiser for it. This is exactly what Agrograde – the brand name under which the company Occipital Technologies Pvt. Ltd. operates – is trying to solve.
According to him, Agrograde is a combination of software and hardware to grade and sort vegetables by quality and size. The company uses computer vision, deep learning and machines to achieve this. It not only ensures that the buyer is aware of the quality and grade of the produce, but also that the farmer gets a good price for the entire consignment. “Our focus is on low value, high volume crops,” adds Kshitij, a mechanical engineering graduate who worked on an industrial automation project before starting Agrograde.
The problem for farmers is that they produce a crop, harvest it, sell it to an aggregator or a trader and move on to the next crop. They do not have the time or the resources to grade and sort the produce by quality and size, because of which they do not get a fair price for their produce. The buyer looks at the lowest quality of the consignment and quite often fixes the price for the entire consignment based on this even though a bulk of the produce may be of a better quality. The buyer too, when he sells either in the market or to a larger player in the ecosystem, does not have the wherewithal to sort the consignment by quality and size and sells it after fixing a margin for himself.
According to Kshitij, they studied the problem statement for nearly nine months by visiting various markets where onions are traded before beginning to work on the solution. It was while working on an industrial automation project that the idea for Agrograde came about. A customer who was into processed foods shed some light on their procurement cycle. There was a regular dispute between the supplier and them over quality resulting in a constant struggle for them. In the industrial automation project, he was working on identifying defects in manufacturing. “For example, a component of an assembly, does it have scratches, is it damaged, is the shape matching the previous one. We used to scan it using computer vision and deep learning,” says Kshitij. So, why not use the same concept to grade and sort fresh agriculture and horticulture produce in terms of quality, size and shape. That is how the vision for Agrograde took shape.
With perishables, as it is you are dealing with a short shelf-life product. And, if it involves shipping to distant markets without proper storage, the chances of more of the produce getting rotten or damaged is higher. However, if the produce is segregated at source itself, the better quality ones in the consignment will fetch a higher price than those that are not so good in quality. If, say, 10 per cent of a consignment of 10 tonnes is of poor quality, the farmer or the trader or the aggregator need not have even shipped it. It would have saved them the logistics cost and also prevented the rest of the consignment getting affected and thus fetching a lower price than what they would have otherwise commanded. Had it been graded and sorted, it would have fetched a different price.
Kshitij co-founded Agrograde in 2018 with Rakesh Barai, whom he got to know while he was working on industrial automation. Rakesh was part of the same incubator in Navi Mumbai – Centre for Incubation and Business Acceleration – as Kshitij. They launched their product in 2019 targetted at farmers, traders, aggregators and farmer producer organisations.
According to him, they photograph every piece of vegetable or fruit using multiple industrial grade cameras. These images are analysed in real-time at speeds as high as 10-12 fruits or vegetables per second. This is the software part of their product. The product is then segregated using a machine they have built which is 7.5 m long and 2.5 m wide. “We have different mechanisms for different crops. It depends on how delicate the produce is, the shape and size and we categorise the produce into multiple grades,” says Kshitij.
He explains how this product of theirs would help build trust and bring visibility in the supply chain. This trade has frequent cases of malpractices, both from the buyer’s and supplier’s side. Farmer Producer Organisations (FPO) usually supply produce to buyers on credit. By the time the produce reaches the buyer’s location, the prices fluctuate. Since quality is subjective in nature and there are no standard tools to measure quality parameters, these unethical buyers use it as a loophole to make money or avoid losses. This malpractice is quite common in perishables as the decision whether to find another buyer or accept a debit note comes with a very short time constraint. A lot of FPOs, traders, and aggregators have faced losses to the tune of ₹20-30 lakhs because of such malpractices. A single debit note wipes out all the profits generated by the previous 5-6 consignments. The current sampling process used for quality checks can be very biased and often results in disputes.
The lack of a standardised solution for quality assaying and grading-sorting also hurts buyers when a vendor mixes disproportionate amount of low grade produce in the consignment. Quality check is only done on 0.5-1 per cent of the produce taken as samples, hence the results can often be misleading. If the grading and sorting are done by a machine that can match the quality specifications provided by the buyer, the scope for cheating is eliminated. This enables trust that would solve the biggest roadblock for digital trade in fresh produce. Kshitij says they have been able to increase discoverability for FPOs by helping buyers source assured quality from them, thus creating a win-win situation by enabling trust and visibility in the supply chain.
Agrograde raised a seed round in July from CIIE.CO, Social Alpha and Villgro, which it used to develop and build the prototypes. According to Kshitij, they built four prototypes before they went commercial. At present, they can grade onions and potatoes and will shortly be capable of grading tomatoes, before they move to other fresh produce.
They have two business models. One is, the trader or an exporter or agri start-ups will buy the machine from Agrograde. The second is, Agrograde will operate the product as a service for anybody who wants it on a pay-per-use basis. “We want to make technology more accessible,” says Kshitij. According to him, they have sold two machines and deployed one on a pay-per-use basis. They can manufacture three machines a month, with a lead time of two months. “Our machine starts from ₹7 lakhs. It is the most affordable optical grading and sorting solution for fruits and vegetables globally,” asserts Kshitij.
The biggest challenge, says Kshitij, is creating an awareness of the technology and ensuring that the benefits reach all players in the value chain – beginning with the farmer and covering all other significant players. (EOM).