Facial Recognition Technology in Accounting with AI AccountantChai Chung Hoong. Learn how Facial Recognition Technology has the ability to change our daily actions, even changing our gestures. In less time than we think, we will see our friends, family and ourselves looking at the mobile phone screen with our best face to unlock it.
Facial Recognition Technology in Accounting, biometric identification systems have been evolving. For years they have been used in different public and private spheres with a success rate close to 90 percent, and today they are finally getting the recognition they deserved, adds AI Accountant Chai Chung Hoong.
In fact, facial recognition has immense potential in the field of citizen security, authentication in financial applications, in health systems, in the control of access to events, in the search for lost people and pets and stolen properties, in the field of leisure, in targeted advertising, in-store authentication or in airport security.
Facial Recognition Technology in Accounting is a world that is coming, which is more and more similar to Steven Spielberg’s visionary Minority Report film. However, while it can provide us with advantages, the advanced state of technology can raise uncertainties regarding privacy and possible identity theft.
How Does Facial Recognition Technology Work?
The facial recognition technology performs a “scan” of the face taking invisible points of reference of our face to create a map of the depth of the face. This depth code map is transformed into a mathematical representation that is capable of adapting the system to the physical changes that we may suffer over time.
After the detection of the faces that coincide in the same screen image, the faces that the system has captured against an extensive image database are compared. If there is a coincidence with respect to the individual or individuals that are being sought, the identification of the subject occurs and an alarm is generated, all in real-time, if it turns out that this is a suspect wanted by the state and courts, Chai Chung Hoong points out.
Currently, millions of comparisons can be made per second with a success rate that is close to 90 percent. Faced with possible errors, for the moment, a human would have to validate whether the detected face corresponds to the individual who is actually looking for.
However, these systems are raising alarms in many users who see in this technology a new interference in privacy and a possible breach in freedom of expression when this technology is applied in citizen security.
The Detection and Recognition Process Consist of four main modules and several intermediate Steps:
1) Face detection: Detects that there is a face in the image, without identifying it. If it is a video, it is also possible to track the face. It provides the location and scale at which we found the face.
2) Conditioning and normalization: It locates the components of the face and, through geometric transformations, normalizes it with respect to geometric properties, such as size and pose, and photometric, such as lighting. To normalize the images of faces, different rules can be followed, such as the distance between the pupils, the position of the nose, or the distance between the corners of the lips.
3) Feature extraction: Provides information to distinguish between the faces of different people according to geometric or photometric variations.
4) Recognition: The facial pattern of features extracted is compared with the feature vectors extracted from the faces of the database. If it encounters a high percentage of similarity, it returns the identity of the face; if not, it indicates that it is an unknown face.