India
October 30, 2020
A little box that can predict the amount of harmful aflatoxin contained in a handful of sample groundnuts… sounds like a far-fetched notion? Not anymore. A collaboration between Pure Scan AI and ICRISAT to create a portable aflatoxin detector has won the Inspire Challenge by the CGIAR Big Data Platform at the recent Big Data Convention, earning a US$ 100,000 grant to build and scale up the device. Utilizing the blacklight fluorescence feature of aflatoxin, this device captures the fluorescence by cameras with filters. Images are processed and the fluorescence degree and pattern are fed into a learning model that predicts the quantity of aflatoxin present in the sample to an accuracy of 1 part per billion error margin.
While more work needs to be done to bring this innovation to the farmer, e.g. an android app and a web platform have to be built, the innovators are hopeful that the device will soon enable farmers to access online marketplaces for a fair price on their high-quality produce free of aflatoxin. For more details on the product, click here for an explainer video on the Rapid Low-Cost Aflatoxin detection using AI.
Aflatoxin – a carcinogenic toxin found in groundnut (and other produce e.g. maize, chillies, rice, various seeds etc.) contaminated by a fungus Aspergillus flavus – can cause liver damage, malnutrition, immune suppression and cancer. Aflatoxin contamination is also responsible for millions of dollars in trade loss for farmers, processors and exporters. At present, there is a dearth of affordable and accessible tests to detect aflatoxin in agricultural produce; also, there is inadequate transparency in sales of these products, making traceability of contaminated products difficult. The above aflatoxin detection device hopes to leverage artificial intelligence and big data to resolve the above challenges, giving farmers a good price for their safe produce.
The Inspire Challenge stimulates CGIAR centers and external partners to link high technology with agriculture and development to deliver impact in vulnerable regions of the world. It encourages participants to leverage digital innovations viz. artificial intelligence, machine learning, robotics, etc. to make life-changing transformations for marginal populations globally.
Other winners of this co mpetition are:
Another innovation co-developed by ICRISAT – Rapid plant disease detection phenotyping – also made it to the 15 finalists in this year’s Challenge. It used hyperspectral imaging and AI for automated early stress detection in chickpea crop, with the potential for large-scale automated stress phenotyping. Click here to see the video: https://youtu.be/OHHfzLIzhPI
The theme for this year’s Big Data Convention (19-23 October) was Digital Dynamism for Adaptive Food Systems, with side events and discussions examining how modern technologies and digital tools could help food systems become more responsive to crises (like the recent global pandemic) that affect food and nutrition security for millions, and more resilient to rebuild themselves quickly and efficiently. Due to restrictions on travel, this year’s Convention was held virtually, with hundreds of delegates logging in from across all the CGIAR centers and their partners around the world. This was the first event that was held under the aegis of
One CGIAR.
For more on our work in digital technology for agriculture, click here: http://exploreit.icrisat.org/profile/knowledge%20management/79
Species
