Gaze Tracking in Liveness Detection | Keeping an Eye on Everyone’s Eyes

William Henry once said that the eyes shout what the lips fear to say. Wherever a person looks and focuses can be recorded through gaze tracking systems which can help behavioral analysis aimed to prevent crimes. Despite the use of Iris and Retina scanning employing human eye scanning, human eye gaze pattern tracking is now considered a next-generation biometric security tool that will not only make it challenging for fraudsters to spoof an eye recognition system but also prompt detection of a potential crime will be enhanced.  

Today, we will discuss how gaze detection works best with liveness detection and its implications. We will also explain some technical concepts related to eye gaze tracking systems and how they can become uncompromised digital security protocols to enhance biometrics.

What is Eye Gaze Tracking?

Eye gaze tracking system monitors and analyzes where and how a person is looking and focusing through the eyes. It uses different sensors, cameras, and other associated technologies to detect the slightest eye movements and determine the point of focus. 

How Eye Gaze Tracking is Different from Iris and Retina Scanning in Biometrics?

Even though eye gaze tracking, Iris, and retina scanning use the human eye for their human recognition goals, eye gaze detection is far different as its goal is to detect and predict human intentions whereas Iris and Retina scanning are used for identity verification.  

Uses of Eye Gaze Tracking

The eye gaze tracking system is very specific in terms of use cases. Unlike Iris and Retina scanning (that also use the human eye for recognition), Eye gaze detection captures the eye movement and predicts the intentions of a human being. We can say that it is mainly used for security and surveillance purposes but gaze tracking can’t be directly and solely used for identity verification. Its major use cases include:

• Eye-controlled navigation in online gaming
• Improved user liveness detection passively i.e. it enhances the user presence detection.
• Assisting impaired individuals through eye gaze and blink detection is used for mobility in devices.
• Fraud prevention and crime investigation (lie detection)
• Deep market research and targeted ads are done by assessing consumer preferences through the way they focus and the products they focus on.

Tech-Talk | Advancing Eye Movement Analysis

There are 2 primary concerns while discussing the technical side of eye-tracking technology. The spatial quality of gaze-tracking technology can be measured in terms of:

1. Accuracy determines the offset between the true gaze point and the recorded position.
2. Precision measures the dispersion of recorded gaze points during the fixation event.

Both accuracy and precision are measured in x and y coordinates as below:

Important Events in the Gaze Detection System

The research focused on remote eye-tracking technology using cameras attached to the device’s screens just like smartphone cameras. This research terms the applications that react to the changing gaze of users as gaze-enabled apps. The most important events in a gaze detection system are listed below:

1. Fixations | A period less than 100 ms up to a few seconds for which gaze is static in a given small region.
2. Saccades | A rapid and ballistic movement between fixations is called saccades. Its normal range is 30 – 80 ms at 2 degrees or larger.

The Factors Influencing Gaze Tracking Quality

• Gaze tracking heavily relies on pupil and corneal reflection and the slightest pupil movements. It is observed that occlusions like heavy mascara, thick frames of glasses, and reflections of light and objects in the eye pupil potentially impair the quality of the captured gaze image. Thus, the quality of gaze tracking is determined by various factors that are related to the properties of the user’s eye, the environment, the calibration procedure, and the gaze tracking methods opted for. 
• Using contact lenses and physical impairments such as squint eyes also affected gaze tracking accuracy and precision.
• Varying physiologies of the eyes, different levels of sunlight, artificial light, resolution, and focus of the camera also impacted the overall gaze detection system’s quality.

The Unseen Bond

Gaze recognition and liveness detection are closely related. Liveness Detection is a basic countermeasure against spoofing attacks that may attempt to bypass biometric security systems in several ways and types. Gaze recognition can be performed on a living human eye and can help in the detection of spoofing attempts through digital means. 

Gaze Tracking | A Super Defender Against Deepfakes

Deepfake injection attacks are advancing and evolving making it difficult for regular anti-spoofing systems to detect them. Biometric identity solution providers are now exploring ways to combine the powers of different biometric traits to build an impenetrable firewall against deepfake attacks. Gaze tracking can play a central heroic role in boosting and detecting deepfake attacks that may take place in a live video call or pre-recorded videos. It will do this by sensing anomalies in the eye movement and gaze patterns which utilizes AI and ML as its core mechanisms.

Despite the deepfake detection, gaze tracking is also believed to prevent:

• Paper mask attacks.
• 3D mask attacks
• Video replay attacks

A proposed algorithm demonstrating the relationship between pupil movement (eye gaze estimation) and liveness detection will help in understanding the use of gaze tracking in liveness detection. 

Let’s have a look at the outcomes of this work:

• 5 important steps were involved in verifying liveness through pupil movement
  • Eye area extraction from live camera vision
  • Eye area tracking using the Kanade-Lucas-Tomasi (KLT) algorithm 
  • Pupil localization for finding a person’s eye direction
  • Activating random LED on the square frame
  • Verifying pupil direction

Results:

• Overall Success Ratio: 89.7% on all databases.
• Excluding Glasses: The success ratio is 94.8%.
• Comparison: Table I compares the proposed eye location method with others.
• Runtime: It takes around 160 milliseconds for a 167×42 resolution color image.
• Algorithm Testing: Achieved a 98% success ratio in detecting live or fake volunteers.

The above discussion proves that gaze tracking is a robust method to enhance liveness detection in a controlled environment. Yet its use in real-time under varied conditions and factors still has room for improvement. In general, gaze tracking is a handy method for liveness detection. 

Incorporating Gaze Tracking in Facial Recognition Security

Since gaze detection is a precise technology that captures even the slightest eye movements and can help in predicting user behavior, it is quite favorable to use gaze tracking to enhance the accuracy and precision of facial recognition technology too. But the general purpose of gaze tracking is far different from identity verification. Once an individual is identified through face recognition, gaze movement analysis can help in expediting the investigation and monitoring activities in runtime. This may help in protecting the public from potential threats and detect crimes well before time. 

Want to experience the world’s fastest face liveness detection solution? Visit Facia