{"id":502,"date":"2021-03-20T09:56:00","date_gmt":"2021-03-20T08:56:00","guid":{"rendered":"http:\/\/innovatrics.vibration.sk\/?p=502"},"modified":"2022-11-08T09:59:43","modified_gmt":"2022-11-08T08:59:43","slug":"how-the-accuracy-of-face-recognition-technology-has-improved-over-time","status":"publish","type":"post","link":"https:\/\/innovatrics.com\/trustreport\/how-the-accuracy-of-face-recognition-technology-has-improved-over-time\/","title":{"rendered":"How the accuracy of facial recognition technology has improved over time"},"content":{"rendered":"\n
\"\"<\/figure>\n\n\n\n

On our way towards a paperless future, Innovatrics has developed seven facial recognition algorithms in the past two years so as to help speed up processes involving facial recognition software. No biometric system is error-free, but year by year the accuracy gets more and more exact. Since 2017, Innovatrics has improved its facial recognition accuracy four times in case of low-quality pictures and 33 times in the case of high-quality ones. <\/strong><\/strong><\/p>\n\n\n\n

\n
<\/div>\n\n\n\n
\n

Dive into this explanatory article and see how face-recognition technology has improved over time, how the accuracy of face recognition gets measured and what Innovatrics\u2019 role has been in the overall development.<\/p>\n\n\n\n

How to measure the development of facial-recognition algorithms on a global scale<\/h2>\n<\/div>\n<\/div>\n\n\n\n
\n
\n
\n\n\n\n

The Innovatrics algorithm has scored top marks in accuracy and speed. These are crucial for large-scale applications such as document verification, criminal investigation and border control. Read more about the November 2020 results here<\/a>. <\/p>\n<\/div>\n\n\n\n

\n

To check the progress, the Face Recognition Vendor Test (FRVT<\/a>) <\/strong>was invented as a benchmark evaluation method of facial-recognition algorithms for verification and identification. Such a test measures the performance of algorithms developed in both commercial and academic communities worldwide to evaluate their competitive advantage. The result is a list of submitted vendors ranked from the lowest to the highest performance.<\/p>\n\n\n\n

To execute FRVT correctly, we have to measure two kinds of rates, both dependable on each other. <\/p>\n\n\n\n

The first is the false reject rate<\/strong> (FRR, also called false non-match rate or FNMR): the instance of a security system failing to verify or identify an authorised person, hence misreading the data and causing a false rejection. It measures the number of \u2018no found matches\u2019 with regards to identical faces.<\/p>\n\n\n\n

To measure FRR, we also have to measure the likelihood of a biometric security system incorrectly accepting an access attempt by an unauthorised user. That is called the false accept rate <\/strong>(FAR, also called false match rate or FMR), meaning the number of incorrectly found matches.<\/p>\n<\/div>\n<\/div>\n\n\n\n

\n
\n
\n\n\n\n

How algorithms have been improving <\/h2>\n\n\n\n

There are several ways to improve the accuracy of facial recognition. The first is “education”<\/strong> \u2013 the better the dataset that the neural network is trained on, the better are the results. To help the machine learn better, the datasets have to be labelled correctly and checked for mistakes. The other avenue is to rely on improving computing power.<\/strong> Neural networks can be more precise and their outputs can be tested repeatedly to find and repair their blind spots.<\/p>\n<\/div>\n\n\n\n

\n

The accuracy of FRVT has to be measured via an FRR (false reject rate) and FAR (false accept rate) ratio; the number of times a camera incorrectly detects a false match (there\u2019s a match found, even if the face is not in the database) to the number of times a camera fails to find a correct match (the camera does not recognise a person\u2019s identity, even though they are already in the database). <\/p>\n\n\n\n

\"\"
In the first case, no match was found despite the fact that the two faces being compared are the same. In the second case, the match was found, even though the two faces being compared are different. In both cases the machine failed to recognise the identities.<\/figcaption><\/figure>\n<\/div>\n<\/div>\n\n\n\n
\n
\n
\n\n\n\n