How to Build a Smarter Access Control Security System for Factories with AI Facial Recognition
< All Articles

How to Build a Smarter Access Control Security System for Factories with AI Facial Recognition


Over the past decade, with the help of rapidly advancing technology, automation has continued to improve to the point where entire assembly lines from start to finish are now almost completely automated. Modern facilities need to evolve and adopt modern automation capabilities to stay relevant amidst today’s continually evolving manufacturing needs.

Many facilities, including large factories, have already begun their journey on the road to digital transformation. Meanwhile, artificial intelligence (AI) technology continues to develop in image recognition, such as with faces, objects, and license plates, improving our daily lives and business environments.

With that in mind, automated access control procedures have become one of companies' primary goals in managing employee security and protecting confidential information. Fulfilling this goal is especially critical for mid-to large-size factories, where hundreds and even thousands of people enter and exit the facilities every day. In addition to employees, contractors, visitors, and foreign technical experts also add to the mix of people to manage in an access control system. With such a variety and complexity in identities, it’s no wonder that many factories and warehouses have gradually adopted facial recognition technology in their initiatives, even integrating attendance tracking systems, to improve access control and identity management.

This article will share the advantages of using facial recognition in factories, its key challenges, and how to enable facial recognition access control for modern factories.

Key Use Cases and Advantages of Facial Recognition Technology

Facial recognition already exists in many access control scenarios we are already familiar with, like Apple’s Face ID, digital and mobile payment solutions, and other scenarios. Since the outset of the Covid-19 pandemic, numerous companies have integrated facial recognition in offices and work environments to facilitate health and safety checks. Companies will still need touchless health screening technology to promote health and safety initiatives and ensure their employees' security as we enter a post-pandemic environment.

Consequently, facial recognition technology will continue to bring value to different industries through system integration, impacting security management, identity verification, and more. Companies can apply facial recognition technology to various scenarios according to their unique needs and requirements. The following is an in-depth introduction to adopting facial recognition technology for large-, medium-, and small-scale environments.

  • Small-scale: offices, commercial buildings

    Since the outbreak of Covid-19, traditional attendance systems of clocking in/out increased the risk of contact with others and, thus, the risk of infection. Luckily, implementing an all-in-one, facial recognition access control solution can manage time and attendance while meeting health and safety requirements. On top of time and attendance management, the newest access control systems can detect mask-wearing and still perform facial recognition, and measure body temperature when equipped with thermal cameras or infrared thermometers.

    In addition, it also dramatically reduces human contact by combining with network IP cameras in commercial buildings for better monitoring. During peak work hours, it increases the efficiency of people flow while reducing time and labor costs usually associated with manual temperature measurement. Finally, it avoids loss and theft of access control cards, misappropriation of identities, and theft of confidential information.

  • Medium-scale: factories, warehouses, cleanrooms

    Many industries, especially semiconductor manufacturers and logistics companies, are undergoing digital transformation. Hundreds of people, including employees, visitors, contractors, and foreign technical personnel, enter factories, warehouses, and clean rooms every day. Stricter entry and exit procedures are necessary to ensure company data is kept secure. In addition, with the development of more intelligent logistic tracking technology, logistics providers now face a greater need to ensure the reliability of their transportation with stricter access control policies at their facilities. Learn more about how manufacturing plants and warehouses have integrated facial recognition.

  • Large-scale: petrochemical and heavy industry plants

    Along with large manufacturing plants, petrochemical and heavy industry plants have acres of outdoor facilities where thousands of people enter and exit every day, sometimes multiple times a day. To effectively understand and control employee access to specific facilities, these companies must manage copious amounts of employee time and attendance records and access privileges.

    Luckily, through facial recognition technology, they can install cameras on large workstations to identify multiple people and record employees’ entry and exit accurately and rapidly. They can also use facial recognition technology to restrict machinery and equipment operations to personnel who have the correct privileges to use that specific piece of equipment.

    In addition, by integrating facial recognition software with existing video management systems (VMS) and real-time messaging systems, companies can also improve the overall efficiency of their personnel supervision. Security personnel no longer need to switch manually among multiple monitor screens. Moreover, monitoring the workflow can improve operations management efficiency and optimize automated processes.

Six Major Challenges Encountered in Large-Scale Environments When Integrating Facial Recognition

Facial recognition is becoming increasingly popular for security uses in large-scale environments, such as factories and plants. Many large-scale factories are exploring introducing AI facial recognition solutions into their existing work environments but must first overcome the six challenges outlined below.

Poor indoor and outdoor lighting

Due to entry and exit issues concerning cross-plant personnel and vehicles, large-scale environments, including petrochemical plants, will need to set up indoor and outdoor cameras and integrate facial recognition technology in their smart access control door systems.

Take an outdoor setting as an example. During the day, faces are severely backlit. At night, images may have excessive noise or appear too dark. Consequently, the harsh conditions of outdoor lighting often lead to inaccuracy in most facial recognition technology.

Fortunately, FaceMe® facial recognition technology uses CyberLink’s unique TrueTheaterTM image optimization technology to sharpen edges, enhance quality, decrease backlighting, and reduce noise, all before performing facial recognition. This technology expedites the recognition process while increasing accuracy by 11.5%.

Partially covered faces

Because large-scale, cross-plant factory workers operate machinery for long hours in outdoor environments while abiding by pandemic prevention and personal safety procedures, they need to wear face masks, protective helmets/hats, and safety goggles the entire day. As a result, their faces are partially occluded, making facial recognition more challenging. In addition, during peak hours of employees clocking in and out, workers generally need to take equipment on and off, reducing efficiency and increasing staffing costs and time spent.

FaceMe’s facial recognition technology employs deep learning to capture facial characteristics and compare them with employee databases. Even when users wear face masks, FaceMe still attains outstanding recognition accuracy: as high as 98.21%. This ability allowed FaceMe® to rank among the world’s top AI facial recognition engines in the NIST FRVT (National Institute of Standards and Technology Face Recognition Vendor Test).

Camera angle limitations

When large-scale environments want to introduce facial recognition technology into network IP cameras or edge devices for entrances, various implementations can sometimes restrict the camera angles and images captured. Contributing factors include an excessive distance between the lens and face, improper backlighting, and too wide of a field of view (FOV) required. These factors may capture only half or even none of the face, leading to the inability to verify personnel accurately.

FaceMe’s facial recognition technology can perform real-time recognition of more than ten people in the same frame. You can even place the camera at heights of 180 to 200 cm and at special angles to capture top, side, and partial views. To perform accurate facial recognition under these restricted conditions, FaceMe® also supports a variety of portable industrial and desktop systems (e.g., Intel NUC), as well as network IP cameras. With such precise technology, FaceMe® achieved a 96.98% recognition rate in the world-renowned NIST FRVT 1:1 WILD test. Furthermore, FaceMe® allows you to conduct facial recognition remotely, providing stricter security control measures even for large-scale scenarios.

High-density people flow

The daily flow of personnel in large factories can be hundreds or even thousands, including contractors, temporary construction workers, and external technical experts. During peak hours, factories need a solution capable of performing facial recognition quickly while providing access control and check-in.

CyberLink’s FaceMe® facial recognition search engine enables GPU acceleration on hardware such as NVIDIA® CUDATM and Intel® MovidiusTM. It performs ultra-fast searches even when databases are massive, taking only 1.6 milliseconds to search for a single face among 6 million.

For more information, read our success story on how we partnered with ACE Biotek to verify identities and measure temperatures in record time.

Use of spoofing technology

Due to the complexity of people inflow and outflow in large-scale environments, companies may encounter access control security issues stemming from spoofing, such as employees clocking in or out for each other, use of face photos to crack access control systems, and unauthorized entry due to loss or theft of an employee’s access card. These issues are especially prevalent in some large-scale petrochemical plants. To reduce personnel expenditures, they have installed edge devices with facial recognition systems in reception areas so that employees can self-confirm their identities before entering secure factory areas.

FaceMe® face search engine supports a range of live recognition and anti-spoofing mechanisms. FaceMe® uses 3D depth cameras or 2D cameras to determine whether a person is real, thus achieving stricter safety controls.

Facial recognition software incompatible with existing hardware devices

When introduced in large-scale environments to strengthen access control security, facial recognition systems may often encounter incompatibility with hardware devices. Many manufacturing and raw materials factories want to import AI facial recognition technology and data/statistics directly from their software and hardware devices (turnstiles with camera lenses, network IP cameras, access control systems for edge devices, video management systems). But because of incompatibility issues in software and hardware settings, installation costs consequently increase.

As a cross-platform facial recognition search engine, FaceMe® supports multiple operating systems (Windows, Linux, Android, iOS). It also supports various machine learning engines such as OpenVINO™, TensorRT, TensorFlow, and CoreML, and can run on personal computers, workstations, and servers. No matter what hardware settings you choose, FaceMe® provides the most suitable setup for each use case.

FaceMe® Security is based on the technology of FaceMe®, the world’s top facial recognition engine. It’s the best choice for large-scale environments looking to implement facial recognition for security control uses.

FaceMe Security: the Preferred Facial Recognition Solution for Large-Scale Access Control

FaceMe® Security is an intelligent solution designed specifically for smart manufacturing security controls. Through real-time face matching, mask and temperature detection, third-party system integration, and other value-added functions, it helps large-scale environments increase the safety of their access control systems and overall efficiency. Read on for a detailed introduction to FaceMe® Security.

FaceMe® Security Add-ons

As we enter our “new normal,” health and safety measures will remain vital for smooth business operations. To most effectively reduce direct person-to-person contact and monitor individuals’ health, FaceMe® Security offers three different add-ons to supplement existing software and hardware. The result is a one-stop solution for access control, health management, and physical and environmental security.

  • Check-in Add-on: This add-on is suitable for lightweight devices with low power consumption. You can also connect to IP cameras to quickly capture personnel’s facial features, compare the person’s identity with your database, and immediately display visitor sign-in or employee attendance records.
  • VMR (Video Management and Recording) Add-on: The add-on simultaneously supports nine cameras that provide real-time monitoring with video recording and playback features. With FaceMe® Security Central, your site’s security management can be impenetrable – security personnel can immediately receive specific event alerts or search recorded video files.
  • Health Add-on: Using facial recognition technology, this add-on supplies a non-contact health monitoring solution to detect mask-wearing status while simultaneously measuring body temperature and verifying identities. You can also send instant notifications to security personnel to take corresponding measures when detecting abnormal body temperatures.

Maximizes facial recognition capabilities with support for multiple edge device types

FaceMe® Security can support various edge computing devices for all environmental sizes. Companies can choose from industrial computers equipped with Intel CPUs/VPUs, NVIDIA Jetson AGX Xaviers, and large-scale Windows workstations with multiple GPUs based on hourly people flow. For instance, a workstation with four NVIDIA RTX A6000 GPUs can support 172 IP cameras. If you need to manage 300 cameras, you only need to set up two workstations.

Flexibly integrates third-party systems, such as VMSs, access control systems, and ERPs

You can also integrate FaceMe® Security with major VMSs (Milestone, Network Optix Nx Witness, VIVOTEK VAST2) and connect with third-party systems such as instant messaging, access control, and ERP. When specific events occur, such as a VIP or blocklisted person entering, the control center can immediately alert security personnel to take corresponding safety measures and protect employees.

For more information, please refer to our FaceMe® Security System Requirements.

FaceMe® Security real-life applications

You can use FaceMe® Security across a wide range of purposes and scenarios. Its infrastructure begins at the front entrance. Edge devices and IP and temperature sensing cameras using FaceMe’s face detection and facial feature extraction functions can measure body temperature and detect mask-wearing. To verify identities, it simultaneously sends facial data through the server and back to the control center to compare against the database. It’s also possible to perform visitor sign-in and employee check-in through the Check-in Add-on. Security personnel can also use the VMR Add-on to view real-time camera images and, when particular events occur, receive warning notifications then take necessary actions.

Final Thoughts

Because of the rapid expansion of automation in large-scale factories and facilities, specific environments such as indoor and outdoor factories, plants, and warehouses can all achieve the following three goals by using FaceMe® Security facial recognition technology:

  • Clock employees in and out, using a safe, touchless method to perform access control, manage time and attendance, and save records for future inquiries
  • Implement health and pandemic prevention actions by detecting mask-wearing status and measuring body temperature, ensuring the safety of employees
  • By installing edge workstations, servers, and VMSs, comprehensively improve security control management of large-scale environments and significantly reduce personnel and operating costs

Retail, restaurant, hospitality, and medical industries have already adopted AI facial recognition technology, but it will undoubtedly find its way into more industrial scenarios. For more information, please refer to our FaceMe® case studies.

FaceMe®: CyberLink’s Complete Facial Recognition Solution

Our Sales Team

For further information on FaceMe®

Our team of experts will be happy to answer your questions and schedule a demo. Free evaluation versions of the FaceMe® are available to qualified contacts.

Contact Our Sales Team

Subscribe to FaceMe® Newsletter