24th October 2022
Technological innovations surprise us every day. Artificial Intelligence enters our everyday life and work. Our professional life is rapidly changing thanks to the introduced technologies, e.g., facial recognition. Is it possible that facial recognition technology won’t let you into the office one day because of a shaved mustache? How can that be helpful or annoying, on the other hand? From this article, you will learn why it is worth implementing Edge AI Face Recognition solutions in the example of specific business situations.
What is face recognition?
Face recognition is a biometric system that identifies people based on the image of their faces. It is based on deep neural networks trained on hundreds of thousands of examples (training data set). They require high computing power, so a standard solution is to use cloud resources, such as AWS or GCP. However, an adequately prepared and optimized model can be run on IoT devices with equally high efficiency and speed, offline – without the need for internet access, which is precisely the goal of Edge AI. When combined with other sensors or components, the input data may contain more information, increasing facial recognition’s effectiveness.
The facial recognition solution prevents identity theft, manages queues, and significantly reduces waiting times. It can also increase the security of your employees. We have many ideas for this system, whether in public transport or commercial buildings. It has many valuable use cases thanks to the most modern technologies – Edge AI.
Are you interested in the Edge AI Face Recognition System?
Will facial recognition let you on the bus?
Edge AI Face Recognition responds to the need for automatic and contactless authentication. Where and how exactly? Let’s try to imagine a multi-million city with public transport used by millions of citizens every day. Rush hour, with gigantic queues at payment gates and overcrowded public transport with rushing citizens. It is hard to imagine ticket controllers who can reliably and precisely check each passenger’s ticket or the correct payment. How to even get to every passenger?
For example, in London in the first quarter of 2022, there were between 6.2-7.0 million people per day using public transport. A challenge in such a crowded place is verifying the reduced-fare tickets. Is it valid? Does the ticket belong to a person who has it physically? Does the discount on the card match the citizen’s data? Imagine a morning in such a metropolis. 6,5 million people. What about contactless and automatic facial recognition gates? It would significantly accelerate and improve functioning, especially in crowded metro or bus lines.
How many times do you check your watch while hurrying to work?
The work-life balance philosophy appeals to us more and more. Many people are looking for a way to organize time for pleasant and relaxing things. Still, endless queues of employees line up every day before their shift. Many workplaces use fingerprint recognition devices for identification. Let’s assume that there are 2,000 employees per shift in the factory. Identifying such a person takes approx. 5 seconds. It takes almost 3 hours to check every single one of them.
The facial recognition system only requires the person to pass through the gate at the designated place. AI model recognizes the face and classifies it based on previously introduced data. It takes about 1 second per person! In this particular case, you can see the time reduction from 2h 43m hours to 33 minutes. The Edge AI facial recognition system surpasses the previous solution. The main benefits are:
- process automation,
- security in the context of sensitive personal data.
Pandemic advantage of Edge AI Face Recognition system
Contactless. The case of the queue at the factory can illustrate how much face recognition is superior to other people’s identification approaches. For example, fingerprint readers need intact, clean skin tissue to read correctly. We cannot assume that people will always have clean hands. The employee’s hands may not be well prepared because of the glue, paint or other materials on the fingers. Correct fingerprint reading may be challenging to perform. In turn, it will generate an extension of time. With many employees, it turns into hours of delay.
And this is where we come in with our idea. We adhere to the limitations and recommendations of contactless solutions that arise from the pandemic. The most significant advantage is that identification at no stage requires direct contact with the equipment from the person passing through the gate.
What about the mustache?
To be recognized by the Face recognition system, you should add photos or videos of a specific person’s face using dedicated app. These photos are processed into vector data and just after that can be deleted – another advantage in terms of anonymity and GDPR. Vectors are mathematical records of the characteristics that make up a point map of the face. They cannot be used to recreate a specific photo. That is why we are talking about the anonymity of the Edge AI Face Recognition solution.
In addition, our solution ensures data security due to the performance of all activities on the Raco Edge AI Gateway device. There is no need to process them in the cloud. The learned AI model recognizes a face based on this vector data. Edge AI facial recognition perfectly copes appearance modifications such as wearing a hat/glasses or changing your hairstyle. With this solution, you don’t have to worry about your mustache. Do you want to change your hairstyle? No worries! New hair will not impress the learned AI model.
Will you cheat the smart one?
Who is watching you? Cameras! There are many types. Which one to choose? Which will help us the most? It is a fundamental question from the point of view of the effectiveness and security of our system. You can choose one of a few options to analyze and recognize the face. The first is the standard RGB camera, like any good quality CCTV camera used to monitor real estate for security purposes. This system is quick to implement and cheap, but not as secure as our second infrared solution.
You can get full autonomy thanks to more advanced technology. Instead of an RGB camera working solo, we can recommend the extension based on the IR camera with an IR illuminator or depth camera. The first option offers good quality and high accuracy. Using AI technology, this camera recognizes a natural, 3D face and illuminates it with infrared light. High efficiency of face classification, even in difficult working conditions such as the shadowed face.
Depth detection camera provides the most accurate measurement. Each sensor zone sends a pulse of infrared light that returns to the sensor. Then the time after which the pulse came back is converted into the distance, giving us depth. This data creates a depth map, which provides the basis for facial identification. The IR camera with IR illuminator reads the face as a figure.
Therefore, any attempt to cheat proposed cameras by substituting a photo or video of another person’s face for recognition seems impossible.
As you can see – when it comes to choosing the right equipment there are several different options. All depends on your needs and requirements, as well as your budget. Creating dedicated solutions is our passion, so we would be happy to discuss your idea for using Edge AI Face Recognition.