Got questions? Find answers here

Data

In the Face Recognition System, the facial parameters of a human face (either from a photo or directly from a camera) are read and converted to digital values. This data is encrypted and stored in a highly secure Microsoft Azure ecosystem which meets all global standards for data privacy and confidentiality. Ref: http://www.azure.microsoft.com - Products b Cognitive Services
No. Facial data cannot be re-assembled to create a face. Once stored, it can only be used for comparison and for the purpose for which it is intended for.
Since face information is stored as a mathematical model data, there is no capability available to export or import face parameters. Therefore, data cannot be misused for purposes other than facial recognition for SmartFace. In the unlikely event of any loss of data, recovery is guaranteed by the Azure platform capability.
Enrolled faces can be removed, but cannot be archived or copied.
Face Model Backup is managed internally within Azure and is not available for external download. Only user defined meta-data stored along with a face model is available for querying via API.
There is no default clean up. Business specific implementations can be built as applicable.
The attendance management system is a cloud based data management module, for the Human Resources department. The data management module is hosted in a server space rented by the customer. The punch in and punch outs are performed via apps installed in a common Android Tab, iPad or from an employeebs own iPhone or Android phone.

Facial data is stored in the Microsoft Azure cloud space and the staff master data is stored in the clientbs own server space. We have these types of transactions in SmartFace.

Action Azure Clientbs Database Result
Staff punch in Valid Valid PI b Punch In registered in Clientbs database
Staff punch out Valid Valid POb Punch Out registered in Clientbs database
Ex staff whose Azure Face data is not yet removed, presents in front of the terminal Valid Invalid PN b An exception transaction registered for later review
A person whose face is registered in Azure may be as a vendor or on a trial basis, but not available in Client DB Valid Invalid AZ b An exception is registered
A person whose face is not registered in Azure, presents themselves. Invalid XXXXX UK b Unknown, Exception registered

In other words, every facial recognition transaction is logged in the client database for reconciliation, analysis and exception handling.

Not really. Each time an employee goes out for a break, he/she can be made to record the status. However, in such cases, the number of transactions will multiply. It all depends on the business rules of the company.

Security

The registered face information is accessible for face match using an API key assigned by the Azure platform at the time of registration for using the Cognitive Services feature. Hence facial data cannot be hacked or copied by any unauthorized software or persons. Images are automatically deleted after processing in Azure. Read Azure compliance and privacy policy here
Yes. The storage of facial data is governed by Microsoftbs strict security policies. See details here. The software built around it also has enough access controls and audit trails, to ensure data security. The server where the data is stored should also follow such high standards and it is usually so if space is rented from a reputed hosting service provider.

Definitely not

Every client gets a unique key when the client is registered in Microsoft Azure. This is done by Microsoft and not by us. This key cannot be used by others. When facial data of employees are registered, these get grouped under this unique key.

Even within a company, one can have grouping based on departments or work-sites if the number of employees is large. However, in case of such a sub-grouping, staff movement from one site to another, may pose an issue. If the number of employees is not that high, one group is enough for a company.

Each clientbs database and the associated code are all completely independent of each other and hence onebs data cannot be seen by another. There is no common code or common data. Thus, the entire system is highly private, secure and confidential.

We shall take care of synchronization of any software upgrades.

This can be allowed or disallowed depending on what a client wants. In any case, buddy punching is not possible.

Quality

Face enrollment is done in two steps. In the first step, the face data is added to the Azure database. The second step is called btrainingb for the Microsoft face-recognition API to be able to recognize a match. Both the steps together take approximately 2-3 seconds. In case a large number of faces need to be enrolled, the first step can be carried out for all the faces. It is then enough to run the btrainingb step only once now, thus saving time.
Generally not so, because the comparison is not based on the image but on facial geometry measured over different parts of the face. To improve accuracy it is recommended that you upload multiple images of the same person.
In our various tests, there have been only occasional failures. In such cases, the enrolled images were without beard but were matched with faces having thick beard. If a person has the habit of having and removing a beard, often, why not have multiple images enrolled b with and without beard.
It does not matter. Even if the glasses are removed when being compared, the comparison passes. It is not necessary to remove the glasses. Care must be taken that the eyes are open and visible without any reflection on the glasses.
Generally, not recommended. However, we have seen that the match passes often but cannot be guaranteed.
It is likely to fail.
It is likely to fail. Better not have any reflection from the glasses.
Most often the match passes
It does not matter. A match always occurs with or without a moustache.
The match can fail if the face is covered to a large extent. However, if the person is touching his chin or forehead and does not cover it much, the match will pass.
It does not matter. In our tests we have seen that unless the phone is covering most part of the face, the match passes.
If the injury is severe or if the face is covered with bandage, yes, the authentication may fail. However, minor cuts or bruises do not have any effect.
It is not recommended that the head or face be hidden.
It is recommended that only one person stand in front of the camera for optimum results.
It is recommended that the person stands still with the face directly in front of the camera. A side pose or a partly hidden face may cause the authentication to fail.
Typically, not more than 2 seconds
If data connectivity fails for some reason, the system has the capability to detect this failure and store the employeebs id and face credentials in our database. Once the connectivity is restored, the face verification is completed.
Facial recognition could fail because of the following reasons b
  • Image quality is poor (image resolution, lighting)
  • No internet connectivity
  • SmartFace license is expired
  • The person is not standing still or not facing the camera
  • The terminal may be off
One must investigate the reason and resolve.
The transaction will be rejected with an error message.
No. The number of employees registered will not impact face recognition speed.

In case of large number of staff and especially in the case of Labour Attendance, yes it is a possibility. The same can happen for Biometric terminals and Smart Card readers also.

SmartFace can be installed in the employees own smart phones and also on Smart Tabs at the entrance.

An employee has to only open the SmartFace App in his/her own smart phone and the punch in or punch out will happen. No need to stand in the queue. If it is a common terminal, the queue will vanish quickly because the time per transaction is hardly 2-3 seconds.

The system is intelligent enough to detect this. There is only one button to punch in and out.
Yes, there is a camera click sound which informs the person that he has pressed the Punch button.
One can detect if the head is covered or not. It may not be possible to check if the head is covered with a helmet or something else.
Yes
Most likely, the recognition will fail.
There should not be any problem in recognition as found out in different circumstances.
It should not pose a problem.
The Azure eco system can also read the gender, emotional feelings (happiness, sadness, anger, contempt, etc.) and the approximate age. Such information is also recorded in SmartFace for further analysis. This information is very useful, especially to obtain honest feedback. However, recently Microsoft has retired these features and hence no longer available.
Yes. The OS Version must be 13.4 and above. (Latest is IOS 17). The full list can be obtained from here.

All the recent versions of Android are likely to support SmartFace. However, there are android phones that have customised User Interface and customised Android. If SmartFace does not work on such a phone, check the Settings for proper permissions for the App and especially the camera.

If the model number is informed to us at info@innovatussystems.com , we shall also verify and if a patch can be provided, we shall do so.

As it is impossible to test the app in thousands of makes and models prevalent now, a common terminal of a make and model where SmartFace has been proven to work, is recommended as a backup.

This is because of the manner in which location co-ordinates are fetched by the phone. This is not a continuous process but happens once in a few minutes. This is a phone feature and not of SmartFace. Usually there will not be much a variation.

SmartFace has been tested on several browsers, both on phones as well on laptops, and found to be working. In case one finds an incompatibility with a specific browser, check the settings in the browser. If still not resolved, please inform us at info@innovatussystems.com.

Recommended browsers: Chrome, Bing, Safari, Brave

If the mask is covering major parts of the face, especially the nose, mouth, lips, part of the chin - the recognition can fail.

Alternatives

Both are good reliable systems in vogue for several years. However, users would have also experienced issues with such systems in terms of reliability, running costs, buddy punching, loss / misuse of cards etc. One must seriously take into account of the fact that a biometric terminal will be touched several hundred times in a day, causing a serious hygiene concern especially in these days of virus scare. Face Recognition system is a step forward that addresses the lacunae in other methods of authentication.
The advantage of SmartFace is that it is completely hardware independent. All one needs is a smart phone or a tablet. A face recognition terminal at the least can cost about USD800.00. A biometric terminal is atleast USD300. One will need several such terminals in a large office.
Yes. Data from SmartFace can be passed on to the main attendance system using special scripts. This will help in gradual replacement of Biometric terminals. Contact us to learn more.

Control

In all cases, the location co-ordinates are also captured and therefore become part of the daily attendance data. Any exception will come to light when the daily report is reviewed.
This can happen even with Biometric terminals and Smart Cards. It all depends on the discipline enforced in the office and onebs own level of honesty.

Additional

While this is technically possible, we do not support this capability at this time.

Miscellaneous

No, not at this point.
No, not at this point.
While this FAQ has focused on Attendance Management, Face Recognition can be useful for a wide range of purposes. A few are mentioned here:
  • Visior management
  • Authenticating a new employee at the time of joining
  • Avoiding impersonation in competitive examinations
  • Senior Care communities
  • Customer relationship management
  • Child day care centers to authenticate the parent(s)
  • Access control to sensitive locations
  • Billing at point of sale outlets
  • Registrations at conferences
Yes. SmartFace can be implemented as an independent system or as a module in an existing system.
Yes it is highly recommended. There can be several reasons that warrant a common terminal as a backup. Some of which are:
  • Someone may have forgotten to bring his / her own smart phone on a day
  • It could be a case of someone losing his/her phone and waiting for a new phone
  • Battery may have died with no charge
  • A person may have an incompatible phone
  • The operating oystem version in the phone may not support the facial recognition app
  • Some phones may have a very low quality camera or a low memory or may not have data connectivity
  • Some phones have customised user interface that prevent the app from working
The number of brands, models, makes are too numerous to test all possible models and hence a common terminal is always recommended as a backup. The common terminal can be a low cost Android Tab or even an iPAD mounted on a stand or on a wall at the entrance. No worries about bottlenecks. While the vast majority will punch in from their own smart phones, the remaining few whose number will be less, will opt for the common terminal. The same terminal can also double up as a visitor management system.
Yes, it is quite possible. If the existing system allows, we can provide a CSV (Comma Separated Values file) of the daily attendance and which can be imported into the existing system. The formats and the modus operandi can be agreed upon before launch.
Yes, it is possible. We can provide the Facial Recognition Module as a library that can be included in your own system. The modus operandi can be agreed during the negotiation.
Innovatus Systems Logo

Still got questions? Contact Us

About Innovatus Systems

We are in the business of providing innovative software solutions that have been of great benefit to our valued customers.

Repeat business has been a testimonial for our quality of work and a customer centric approach.

Please visit www.innovatussystems.com for more details about us.

Our motto:  Be Innovative. Stay Ahead.

Top