AI Case Study
Simprints predicts what a baby's fingerprints will look like when they reach adulthood through machine learning
Simprints is aiding people whose birth was not officially registered by developing a remote fingerprint scanner. The device called Vero transmits fingerprints to mobile devices via bluetooth. The system, currently in use in Western Nepal enables non-profits and governments to provide identification and verification documents around the world. Machine learning is used on neonatal prints to track babies into adulthood and predict how fingertips will change as they grow. Currently, Simprints is involved in seven projects in mostly less developed countries.
Public And Social Sector
"Based in Cambridge, in the United Kingdom, Simprints has developed a patented, low-cost, low-fuss fingerprint scanner for the company’s biometric data collection and transmission service.
The fingerprint scanner at work in Western Nepal is Simprints’ flagship device. Simprints does not call itself as a hardware company, however. Simprints is an information service. It provides identity documentation and verification methods to non-profits, governments and others working to deliver services around the world.
The device, called Vero, is thus far unique in the field of biometric devices. It is designed for the unusually austere conditions found in the world’s remote and underserved communities.
“The scanner needs to be rugged. Not just airport security rugged — sub-Saharan Africa rugged. It needs to be dust proof, water resistant, lightweight, long battery life, and easy to use,” says Tristam Norman, who co-founded Simprints’ and serves as its chief technology officer.
The scanner weighs 105 grams and is 108mm by 46mm by 41mm — roughly the size of a feature phone. Drop tests find the scanner is shock resistant up to 1.2 meters onto concrete. It can operate in temperatures ranging from freezing to 45°C, at up to 90 percent humidity. It can withstand storage temperatures of -20°C to 60°C. It has an IP rating of 65, signifying complete protection from dust, and resistance to low-pressure water jets from any direction. Tests found no water damage after spraying the device with a hose from three meters away at a pressure of 30 kilopascals.
The scanner runs on a 1050 mAH lithium ion battery that burns out in 4.5 hours under heavy usage (4,000 scans) or 14 hours under lighter usage (200 scans). It charges with a standard micro-USB, which, if lost or damaged, is now relatively easy to find anywhere.
Simprints’ scanner takes 300- by 400-pixel images at a resolution of 500 DPI. The data it produces conform to specifications for biometric data interchange formats by the International Standards Organization and the U.S. National Institute of Standards and Technology. Vero’s data is also certified by the U.S. Federal Bureau of Investigation.
Vero securely transmits fingerprints and analysis via Bluetooth to mobile devices using Android operating systems. The scanner is compatible with several mobile data gathering platforms that healthcare workers, governments, and global development practitioners use, including CommCare, Open Data Kit, and District Health Information System 2.
Simprints’ first year closed after hundreds of hours of field tests and three redesigns of the prototype. In its second year, 2015, the organization won a grant from Innovate U.K. that funded software development and supported pre-production of the fingerprint scanner. The company unveiled the scanner in its current iteration in 2016.
In 2017, Simprints won a $2 million Saving Lives at Birth “transition to scale” grant. The money will fund work with BRAC to deliver maternal healthcare to nearly 4.5 million pregnant women and their children in Bangladesh.
At the time of writing, Simprints is involved in seven projects in Nepal, Bangladesh, Afghanistan, Uganda, Nigeria, Zambia and Kenya. Ten more are on the docket for 2018, Manhart says."
Pilot; Results not yet available
"The technical challenge in analyzing an infant’s fingerprints is predicting how the prints will change as the baby grows. The Simprints team is using machine learning on a dataset of neonatal prints that may be able to track a baby’s prints into adulthood."
R And D
Core Research And Development
"Only 71 percent of the global population has a registered birth, the U.N. estimates. As expected, those who do are distributed unevenly. In sub-Saharan Africa, for instance, only 46 percent of children under the age of five have a registered birth."