As one of the new technologies for identity verification, face recognition is increasingly accepted and used, and the face-opening and withdrawal of accounts, the identity verification of passenger stations, the use of Haixin identity verification equipment in judicial examinations, etc... can be seen Face recognition has been applied in anti-terrorism, pension, finance, taxation, education and other industries. Although the technology is getting more and more mature, the most worrying thing is the security issue, especially when it comes to the vital interests such as money, it is always not safe to use.
What kind of technology is face recognition technology? In terms of security, is it not reliable? In fact, many times, we don't have a deep understanding of face recognition technology. Many times we take it for granted. Let's talk about the common misunderstandings about face recognition and see if the facts are what we think.
Accuracy of face recognition
If the accuracy of face recognition is 98.3%, is that there is a 1.7% error rate? So the liar has a loophole to drill?
Do you think so too? Actually not!
Face recognition accuracy is measured by two indicators: misrecognition rate and rejection rate (pass rate)!
Misunderstanding refers to the fact that it was not originally, but the system recognized him as himself. The misrecognition rate is used to determine the ability of the algorithm to identify the scammer, the smaller the better. At present, the misrecognition rate index commonly used in practical systems is one thousandth or even one ten thousandth.
The pass rate is used to determine the ability of the algorithm to accurately determine a good person. The bigger the better. Pass rate = 1 - rejection rate. That is, the smaller the rejection rate, the better. The pass rate of practical systems is often above 95%.
Under a specific algorithm, a threshold needs to be set as a criterion for determining similarity. When the similarity is greater than this threshold, it indicates that the matching is successful, otherwise it indicates that the matching fails. The higher the set threshold, the lower the false positive rate and the lower the pass rate.
Simply speaking the accuracy rate is not appropriate. The industry commonly used is the pass rate or rejection rate of one thousandth of the misunderstanding rate to describe the accuracy of the algorithm.
Interference with face recognition, such as make-up, easy-to-lerance, facelift, and head
If I make up and dress up, isn't face recognition not working? If I am interested in turning my head, can I still recognize my successful identity verification through face recognition? Are these methods easy to fool face recognition?
The face recognition system mainly extracts the contour key points of the main organ parts based on the face, and extracts statistical features near the key points. These methods make it easier to increase the rejection rate, which means that the face recognition system cannot be identified as a person. But relying on these methods alone, it is difficult to identify a person as another specific person.
Face recognition technology is based on the recognition of facial images. Face shaping may cause certain changes in the face by modifying the bones, but the impact is limited.
If a heavy makeup is made when the photo is registered, it is easy to cover up the facial features, which may increase misunderstanding and increase rejection. Similar to glasses, if you wear glasses, it will block the facial features, especially the feature points around the eyes. These features are relatively heavy in the recognition process and may cause the system to be unrecognizable.
Compared with other biometric technologies, face recognition technology does not need to be artificially coordinated. It is non-contact and can perform identity verification without awareness. The prospects are still there.
What kind of technology is face recognition technology? In terms of security, is it not reliable? In fact, many times, we don't have a deep understanding of face recognition technology. Many times we take it for granted. Let's talk about the common misunderstandings about face recognition and see if the facts are what we think.
Accuracy of face recognition
If the accuracy of face recognition is 98.3%, is that there is a 1.7% error rate? So the liar has a loophole to drill?
Do you think so too? Actually not!
Face recognition accuracy is measured by two indicators: misrecognition rate and rejection rate (pass rate)!
Misunderstanding refers to the fact that it was not originally, but the system recognized him as himself. The misrecognition rate is used to determine the ability of the algorithm to identify the scammer, the smaller the better. At present, the misrecognition rate index commonly used in practical systems is one thousandth or even one ten thousandth.
The pass rate is used to determine the ability of the algorithm to accurately determine a good person. The bigger the better. Pass rate = 1 - rejection rate. That is, the smaller the rejection rate, the better. The pass rate of practical systems is often above 95%.
Under a specific algorithm, a threshold needs to be set as a criterion for determining similarity. When the similarity is greater than this threshold, it indicates that the matching is successful, otherwise it indicates that the matching fails. The higher the set threshold, the lower the false positive rate and the lower the pass rate.
Simply speaking the accuracy rate is not appropriate. The industry commonly used is the pass rate or rejection rate of one thousandth of the misunderstanding rate to describe the accuracy of the algorithm.
Interference with face recognition, such as make-up, easy-to-lerance, facelift, and head
If I make up and dress up, isn't face recognition not working? If I am interested in turning my head, can I still recognize my successful identity verification through face recognition? Are these methods easy to fool face recognition?
The face recognition system mainly extracts the contour key points of the main organ parts based on the face, and extracts statistical features near the key points. These methods make it easier to increase the rejection rate, which means that the face recognition system cannot be identified as a person. But relying on these methods alone, it is difficult to identify a person as another specific person.
Face recognition technology is based on the recognition of facial images. Face shaping may cause certain changes in the face by modifying the bones, but the impact is limited.
If a heavy makeup is made when the photo is registered, it is easy to cover up the facial features, which may increase misunderstanding and increase rejection. Similar to glasses, if you wear glasses, it will block the facial features, especially the feature points around the eyes. These features are relatively heavy in the recognition process and may cause the system to be unrecognizable.
Compared with other biometric technologies, face recognition technology does not need to be artificially coordinated. It is non-contact and can perform identity verification without awareness. The prospects are still there.
Hole Openers,HDD Rock Reamer Hole Opener,Hole Openers For HDD,PDC Hole Opener
CANGZHOU GREAT DRILL BITS CO., LTD , https://www.pdcbits.nl