VisionLabs, a resident company of the Skolkovo Foundation, which specializes in computer vision and machine learning systems, has been ranked as one of top 3 face recognition algorithms of the US-based National Institute of Standards and Technology (NIST).
The company took one of top places in the Mugshot category (images with varied lighting and background and bad image quality), as well as one of top three companies in the VISA category (studio photos in good quality against the white background).
The contest assessed more than 100 algorithms by different suppliers. The criteria for winning projects included precision and speed of data searching and the flexibility of the algorithm in the process of adapting to subsequent changes.
“We at VisionLabs are very proud of the fact that our algorithm has once again been recognized as the most efficient and accurate,” Alexander Khanin, CEO of VisionLabs, told Sk.ru. “We have been working continuously on improving our product so that we could always provide our customers with the highest quality technology. Using our technologies, customers in various fields are able to provide the most efficient and safe work processes that make daily life more comfortable for our customers as well as for our end users,” he said.
Previously the VisionLabs technology had taken top positions in various independent tests conducted by NIST. The technology has proven to be many times more efficient in terms of how much memory capabilities it needed, and its productiveness has also been considerably higher than that of other developers. The recognition error rate of VisionLabs is on average 0.002%, which is again the best figure in the industry.
The table presents the most effective 1:1 algorithms measured in accordance with the False Non-Match Rate (FNMR) method in several different sets of data.
NIS launched the new technology of face recognition in February 2017. The FRVT 1:1 test corresponds to the script of the person’s verification by his or her photograph. This research also seeks to assist the Department of Commerce of the United States to find the best suppliers of such software solutions.
In March VisionLabs took the first place at the largest international competition ChaLearn Face Anti-Spoofing Attack Detection Challenge which took place at the CVPR 2019 conference, the main annual event on computer vision. At the event the VisionLabs team presented its Liveness technology, which was 1.5 times better than the technology that in end took the second place. There were a total of 25 teams from different countries taking part in the final leg of the tournament.
LUNA Face Recognition Platform is the flagship project of VisionLabs Company. It is a system of measuring the biometric data. The platform is being used in more than 40 banks and national credit bureaus in Russia and other CIS countries. The LUNA software complex uses the innovative image recognition technology.