Facial Recognition Attendance Checking System

author
By TNI Tech

Jan 02, 2025

Advanced Attendance checking with AI camera helps to reduce the timekeeping time and effort.

image

1. Application description

Currently, most businesses still use fingerprint machines to control the entry and exit time of employees. Despite its popularity, this solution reveals many disadvantages. Fingerprint machines often take a long time to confirm, especially when there are a large number of employees coming in and out in the same amount of time. In addition, fingerprint technology is prone to inaccuracies, and cannot support reviewing images/videos to post-check if there are discrepancies in attendance data.

tni logo
tni logo

Some businesses have turned to facial recognition tablets to overcome the limitations of fingerprint machines. This solution helps speed up the timekeeping process, minimizing waiting time. However, tablets can only record attendance for one person at a time, causing inconvenience in crowded environments, especially in units with a large number of employees. Moreover, tablet systems often do not have built-in video or image storage to allow post-inspection when necessary.

The timekeeping solution using facial recognition technology via AI Camera has been offered, providing a more advanced, efficient, and reliable timekeeping experience. With AI cameras, the system is able to identify 5-10 people at the same time in one frame, helping to minimize congestion at timekeeping points. In addition, thanks to the ability to store on a recorder, computer, or cloud storage, the system allows for image and video review to ensure absolute accuracy. Not only timekeeping equipment, AI cameras can also be used for other security monitoring purposes, improving investment value for businesses.

tni logo

2. System architecture

There are two approaches to deploy the AI-camera attendance checking system based on the infrastructure of the customer:

  • Customers who have the CCTV system already, we propose to integrate the AI camera software to the existed system.
  • Otherwise, we recommend customer should equip with an all-in-one AI camera system for cost optimization, camera compatibility, unified UI platform, and future scalability.

2.1. Existed system integration

image

The first part of the illustration shows a combination of a conventional CCTV camera and a computer capable of AI analysis. When the camera is connected to the AI analysis computer, it becomes an AI camera, which not only has the function of observation, but also can recognize faces, analyze behavior, and detect abnormal events. The AI camera is capable of processing data directly from the image, allowing for automated tasks and issuing notifications when necessary situations are detected. This helps to optimize surveillance activities and improve the efficiency of security management.

The second part of the illustration shows the benefits of combining a conventional CCTV camera with a Network Video Recorder (NVR). When the data from the camera is stored in the recorder, all images and videos will be stored permanently, serving for the purpose of re-checking when necessary. This feature is especially useful in cases where it is necessary to re-verify information or verify that something has happened. Data storage on the recorder ensures that all activities are fully recorded and easily retrievable to support post-inspection or information collation.

2.2. New-built system

image

Similar to the Existed system integration approach, the intelligent component of this system is the AI analysis computer. However, unlike traditional setups, this computer also runs VMS (Video Management System) software for camera management and recording, eliminating the need for a separate NVR. This integration creates an all-in-one software solution, allowing users to access, manage, and utilize the system through a single unified UI platform.


3. Camera setup

3.1. Camera specification

  • Recommended resolution: 4MP or higher.
  • Frame rate per second: 20-30 FPS, 25 FPS recommended.
  • Video compression standard: H264 or H265, H265 recommended.

3.2. Installation height

The ideal height for installing a camera is usually between 2m and 3.5m from the ground. This is the height at which the camera can record the face of a person at an average height without causing distortion.

Depending on the actual situation, to determine the installation height accordingly. Avoid mounting too high or too low, as this will reduce the accuracy of identification. Setting the camera too high can lead to a suboptimal viewing angle for the face, and too low will be easily obstructed by objects or pedestrians.

3.3. Tilt angle and installation direction

The camera should be installed with an angle of inclination of 15 to 40 degrees from the horizontal line. This makes it possible for the camera to cover the entire viewing area and easily record images of faces at different angles.

It is recommended to point the camera directly in the direction where the person will be moving, e.g. the door or aisle area. Placing the camera perpendicular or slightly tilted towards the observer will increase the facial recognition accuracy.

3.4. Lighting

The camera needs to be installed in areas with stable lighting. Natural or artificial light can be used, as long as it's not too bright or too weak.

Avoid facing the camera directly from a strong light source (such as a large window during the day) as this can cause glare and blur the face.

Use evenly distributed light in the area, avoiding dark areas or sudden changes in light. White light LEDs can be added to facilitate even illumination.

3.5. Distance from the camera to the subject

The ideal distance from the camera to the location where the face appears clearly is usually around 1-3 meters. This distance allows the camera to capture enough facial detail for accurate recognition.

Make sure the area where the camera is installed is not too far or too close to a person's face, to avoid losing important details when processing images.

3.6. Obstructions

Make sure there are no obstructions such as trees, objects, or barriers between the camera and the facial recognition area. These obstructions will make it difficult to identify and reduce system performance.

3.7. Check again after installation

After installation, it is recommended to conduct an inspection to ensure that the camera has been installed at the right angle and location. Gentle adjustments can be made to optimize viewing angles if needed.

Test the recognition system to confirm that the camera captures the face clearly and is not affected by lighting elements or viewing angles.


4. Guidance for timekeepers

For the attendance process to be quick and accurate, please follow these steps:

4.1. Position

In front of the timekeeping camera, keep an appropriate distance according to the instructions (usually from 1-3 meters).

Make sure the face is in the camera frame, facing the camera so that the system can easily identify it.

4.2. Posture and expressions

Keep your face clear and unobscured. Avoid wearing sunglasses, masks, or face coverings.

Make sure the light on the face is enough so that the system can recognize it well, not standing where there is a shadow on the face.

Relax and keep your natural expressions, don't smile too loudly or frown, which helps the system to recognize more accurately.

4.3. Examples of valid cases

Image 1Image 1Image 1

4.4. Examples of invalid cases

Image 1Image 1Image 1

5. Conclusion

The AI Camera-based Face Attendance Checking System offers a modern, efficient, and scalable solution for businesses aiming to streamline their attendance processes and enhance security management. By addressing the limitations of traditional fingerprint machines and tablet-based systems, this solution delivers speed, accuracy, and reliability.

With its ability to recognize multiple faces simultaneously, store images and videos for post-inspection, and integrate seamlessly with existing or new infrastructure, the system proves to be a cost-effective and versatile investment. Its deployment options cater to diverse customer needs, ensuring compatibility and scalability for future requirements.

When implemented with proper setup, including optimal camera positioning, lighting, and system calibration, the AI camera system achieves peak performance. This not only saves time and resources but also contributes to a smarter and more efficient workplace environment. As businesses continue to embrace AI-driven technologies, systems like this will play a pivotal role in redefining operational efficiency and security standards.


6. Case Studies

  • Co May Dormitory: AI camera system helps Co May Dormitory in student management and theft prevention. [Read more]