In June 2023 the European Space Agency began a demonstration, working with the German firm Marple and the non-profit Global Organic Textile Standard, to test whether satellite imagery read by artificial intelligence can tell organic cotton fields from conventional ones across India. It is a modest attempt to answer an old question of trust in the textile trade: when a label says a fiber was grown without synthetic chemicals, can anyone see the field behind it? The effort is being made because the scale of fraud, and even the scale of organic farming itself, remains largely unknown, and the ans
ESA Backs AI Satellite Project to Police Organic Cotton Fraud in India
No one really knows how far fraud has preyed on organic cotton
In June 2023 the European Space Agency began a demonstration, working with the German firm Marple and the non-profit Global Organic Textile Standard, to test whether satellite imagery read by artificial intelligence can tell organic cotton fields from conventional ones across India. It is a modest attempt to answer an old question of trust in the textile trade: when a label says a fiber was grown without synthetic chemicals, can anyone see the field behind it? The effort is being made because the scale of fraud, and even the scale of organic farming itself, remains largely unknown, and the answer may depend on whether machines can learn to read India's small, mixed and varied farms from above.
On June 6, the European Space Agency launched a demonstration aimed at a persistent problem in the textile trade: verifying whether cotton sold as organic in India was really grown that way. Its bet is artificial intelligence reading satellite imagery, built with German firm Marple and the Global Organic Textile Standard, a non-profit behind a voluntary global standard for the industry.
The software will be trained on imagery mainly from ESA's Sentinel-2 satellites. Its task is to find cotton fields across India and sort them as organic or conventional, drawing on vegetation, water and soil signals and indices such as NDVI, which gauges plant health and density. In a prior pilot in Uzbekistan, the system reportedly reached 98% accuracy. India is a harder test, with varied climates, small fields and intercropping that make organic cotton difficult to distinguish.
GOTS joined because it admits how little it knows. There is no reliable count of organic cotton farms in India, and no one can say how deeply fraud has affected the sector. The project is designed to run in two directions: flagging certified fields that seem to fall short, so questions arise before harvest, and identifying traditional, ecologically sound farms, especially small ones hidden from view, that might qualify for certification.
Funded at about 500,000 euros through ESA's BASS program with GOTS, the effort expects first Indian results before year-end. The central question remains open: whether Uzbekistan's accuracy holds across India's fragmented landscape, and whether the farms flagged from orbit prove to be real cases for investigation.
On June 6, the European Space Agency started a demonstration project aimed at a stubborn problem in the textile trade: how to tell whether cotton sold as organic in India was grown that way. The tool it is betting on is artificial intelligence reading satellite imagery.
The work centers on software from Marple, a German technology company. Over the course of the project, the software will be trained on imagery from ESA satellites, chiefly the two polar-orbiting Sentinel-2 spacecraft. Its job is to detect cotton fields across India and automatically sort them by cultivation method, organic or conventional. It is designed to draw on a range of sensors that capture information about vegetation, water and soil. It also uses indices such as the Normalized Difference Vegetation Index, or NDVI, which measures the health and density of plant cover.
This is not the software's first outing. Two years ago Marple piloted the capability with ESA in Uzbekistan, where the company and the agency said it distinguished organic from conventional cotton with 98% accuracy. India is a harder test. Jeffrey Thimm, a project manager at the Global Organic Textile Standard (GOTS), said climatic conditions there are more varied, and the prevalence of small fields and intercropping makes organic cotton more difficult to pick out. Training the AI in those conditions, he said, is important for improving its accuracy.
GOTS, a non-profit behind a voluntary global standard for the industry, is the project's partner. The organization has a stake in the outcome because of what it says it does not know. Thimm said no one really knows the degree to which fraudulent business practices have preyed on the organic sector. He added that there is no reliable data source on how many organic cotton farms exist in India, which makes it hard to say how much organic cotton is actually being grown, or where.
The project is meant to work in two directions. Fields already certified as organic that appear not to meet the criteria would be flagged for investigation before their cotton is harvested, so questions could be raised before the fiber enters the supply chain. In the other direction, GOTS says the system could identify fields farmed with traditional and ecologically friendly methods that might qualify for organic certification, particularly smaller farms operating under the radar.
The funding comes from ESA's Business Applications and Space Solutions program, known as BASS, which is co-financing the effort with GOTS. Together they have put in around 500,000 euros, about $535,000.
The first results from India are expected before the end of the year. GOTS plans to use them to improve yield estimates. The open question is whether the accuracy reported in Uzbekistan holds up across India's smaller, more mixed and more varied fields, and whether flagged farms turn out to be genuine cases for investigation.
Citas Notables
Part of the problem is that no one really knows the degree to which fraudulent business practices have preyed upon the organic sector.— Jeffrey Thimm, GOTS project manager
Quieres la nota completa?
Lee el original en SpaceNews ↗
Technology