Government may go high-tech to reduce wastage of fresh fruits, veggies

The government is exploring a technology tool developed by ecommerce major Amazon called Johari to reduce wastage of fresh fruits and vegetables at its warehouses and retail outlets.A pilot project at some of Mother Dairy-owned Safal stores across Delhi and Bengaluru is on the anvil.

Elevate Your Tech Prowess with High-Value Skill Courses

Offering College Course Website
IIM Lucknow IIML Executive Programme in FinTech, Banking & Applied Risk Management Visit
Northwestern University Kellogg Post Graduate Certificate in Product Management Visit
MIT MIT Technology Leadership and Innovation Visit

Johari uses computer vision models and wi-fi-enabled internet-of-things (IoT) cameras to identify pre-determined defects—cuts, cracks and other damages—in fresh produce.

These cameras are strategically placed on store shelves or sites where produce is stored. They automatically capture images of the crates at periodic intervals, allowing for continuous monitoring of the quality of the fresh produce.”The government is always open to new ideas. We will do the cost analysis and ensure that the technology is scalable as well as viable for the government, based on which we can do a pilot in select outlets,” a senior government official told ET, requesting anonymity.

This development comes after a recent closed-door meeting between officials of government policy think-tank NITI Aayog and Amazon to discuss the efficacy of Johari.

Discover the stories of your interest

The Centre mostly procures foodgrains to be sold at subsidised rates and to maintain a buffer. Besides, it procures onions in large quantities and at times tomatoes to keep prices in check and maintain a demand-supply balance. It also sells fresh fruits, vegetables and other perishables through Safal stores under the National Dairy Development Board.Rajeev Rastogi, vice president, machine learning at Amazon, confirmed to ET the meeting held with NITI Aayog officials to get Johari deployed in government-owned centres.

In August this year, Amazon India launched the shelf monitoring solution Johari to ensure that only fresh quality fruits and vegetables are delivered to customers.

Automated monitoring using IoT cameras reduces the need for manual inspection, saving time and labour costs in the quality control process, the company said. The technology provides real-time information, enabling prompt decision-making and action to maintain the freshness of the produce, it said.

Cost of deployment

The deployment cost of the shelf-life monitoring technology encompasses several factors. To optimise image viewing on each shelf, the recommended approach entails deploying one low-cost IoT camera per shelf, ensuring a practical and effective monitoring solution that is scalable, Rastogi said.

According to him, the selected IoT cameras are priced at approximately $25 each for 5 megapixel and $20 for 3 megapixel, making them highly cost-effective for deployment at scale.

The estimated cost for deploying the camera in each store can range from Rs 2,600 to Rs 3,300 per camera, encompassing the cost of the IoT cameras with nearly 50% import duties, he said. There are other costs for required cloud computing infrastructure and usage of quality models as a service depending on the scale and usage, he added.

The technology identifies the region of the defect on the product surface area and can also calculate dimensions such as ratio of the length/area of the defect with respect to the length/area of the item, he explained.

The total deployment cost is contingent on the number of Safal stores in each region and the number of shelves required to be monitored, he clarified. Each shelf will require one camera for monitoring assuming a standard crate size.

The upfront deployment costs are offset by the technology’s capability to mitigate food waste, enhance quality assurance and elevate overall customer satisfaction, he argued. Johari works in two monitoring modes—manual and automated. Manual monitoring lets Amazon India’s sellers or operators submit a picture of the produce in a crate from their phone.

The shelf monitoring solution then analyses the image using grading logic to highlight any item that does not meet quality standards, Rastogi explained. He said that the app can also be used to understand specific defects for each item and the whole process takes six seconds. In the automated monitoring option, cameras on shelves take pictures at predetermined intervals and analyse quality through the above process, he said.

Two types of machine learning models, trained using annotated defects in millions of produce images, have been developed: one detects each visible item in the crate and counts the total number of items while the second identifies visible defect classes present in each item along with the precise location of the defect and its size (area and length).

Rastogi added that Johari has been deployed only in India and not globally.

Categories: