dida Logo

Smart Access Control with Facial Recognition

We developed a multi-level security system with facial recognition for automatic access control.

Input:Camera images of a person
Output:    Decision if the person is authorized or not
Goal:Provide a reliable, easy-to-use access control system

Starting Point

Our customer is a technology company that offers security systems, among other things. The company approached us with the challenge of developing an automatic facial recognition solution for an intelligent access restriction system. The system should be easy and reliable for companies to use in order to grant authorized persons access as easily as possible and to reject unauthorized persons. 

The authorization process has two steps. First, authorized persons have a radio key that is detected by the system. Second, their face is captured by two cameras when they stand in front of the secured entrance. These images are compared with the images of authorized persons stored in a database. This comparison of the camera images with the stored images is the core of the computer vision solution we developed for the customer.


There were two main challenges:

  1. Establishing identity. The system has to recognize whether the face in the images sent by the cameras corresponds to a face in the database. The challenge is that people may look different on different days, look at the camera differently, have different accessories on, and the lighting may be different.

  2. Verification of authenticity. The system needs to ensure that the images from the cameras actually show a three-dimensional face, and that an unauthorized person is not perhaps just holding up a photo of an authorized person to the camera to gain access.


For the first challenge, we trained a neural network in two steps:

  1. In the first step, we performed training on a very large, publicly available image dataset from the internet. 

  2. In the second step, we refined the model by training with images of volunteers that were taken with the customer's camera system.

As a result, the trained neural network learned by itself which features are important for determining the similarity of faces. It can now handle faces it hasn't seen before. So for a new company that wants to use face recognition, the neural network does not need to be retrained. All that is needed is to use the camera system to capture images of people who are authorized to enter and store them in the database. If they are then standing in front of the door, the system will recognize them.

For solving the second challenge, preventing unauthorized access by showing photos of authorized people, we chose classical computer vision methods. This approach takes advantage of the fact that a person's face is captured by two cameras each looking at the face from different angles. In simple terms, this method works by measuring the distances between certain points on the face. From the differences between the two images, one can conclude whether the face is real, i.e. three-dimensional, or whether it is a flat photo of a person.

Philipp Jackmuth

Managing Director


Tell us a little about your project requirements and we'll get in touch with you.

Further Projects

A selection of projects we have done

Enpal logo

Automatic Planning of Solar Systems

Creative solutions enabled us to automate the process of planning solar systems.

Machine learning arranged these solar panels
Rainforest with waterfall
ESA (European Space Agency) logo

Artisanal and Small Mine Detection

See how machine learning can be used to stop environmental destruction.

Semiconductors in production
Ferdinand Braun Institut logo

Defect Detection in Manufacturing

AI-supported optical defect detection for semiconductor laser production.

Urban growth of a city
ESA (European Space Agency) logo

Monitoring Urban Growth and Change

An image segmentation algorithm that supports sustainable city planning.