Sightcorp logo

Gaze Tracking Technology

Everything about Gaze Tracking Technology

     

What is Gaze Tracking Technology and how does it work?

Gaze tracking technology consists of various components that work together to track where people’s visual attention is concentrated. These components include cameras or infrared-based eye trackers and gaze tracking software.

Cameras capture images of eyes and send this data through to a gaze tracking software program that processes the data and provides useful information about what exactly people are looking at, how long they are looking at it, and in what order they are looking at it.

Gaze tracking software uses image processing algorithms (often based on AI) to interpret the data that it receives from the cameras. The metrics that are provided by the software may include gaze points, fixations, fixation sequences, and areas of interest.

Discover Sightcorp’s Gaze Tracking Solution

What can you do with Gaze Tracking Technology?

Gaze tracking technology enables you to determine:

  • Attention time: how much time someone spent looking at a particular point of interest
  • Attention sequences: in what order someone looked at various points of interest
  • Recurring attention: whether someone looked at the same point of interest more than once

 

You can use this information for many purposes. In the context of user experience (UX) design and web usability testing, for example, you can use gaze tracking to determine:

  • Which parts of a web page received the most attention
  • In which sequences people looked at the content on a web page
  • Which parts of a web page were ignored
  • Which parts of a web page are a potential source of confusion

 

With this information, you can then make important UX decisions, such as where to place call to actions for optimal visibility and clickthrough, where to place the most important information, and how to structure the website menu for easy navigation.

What are some other uses of Gaze Tracking Technology?

Many industries are using gaze tracking technology to improve products, services, and experiences. Here are a few examples from different industries:

Healthcare:
-To conduct psychological tests and to evaluate patients for neurological conditions

Human-computer interaction:
-To enable people to control computers by using their eyes, instead of using a keyboard or mouse (also in the context of gaming)
-To develop virtual reality (VR) and augmented reality (AR) experiences

Accessibility and mobility:
-To enable people to control devices and appliances using their eyes (e.g. wheelchairs, televisions)

Automotive:
-To enhance safety on the roads by alerting drivers when they divert their attention from the road for too long

Market research and advertising:
-To optimize packaging design, store layout, and digital signage/displays
-To optimize the layout and design of print and online ads

Gaze Tracking using Deep Learning

While using regular cameras instead of specialized infrared trackers makes gaze tracking technology more accessible, it also presents certain limitations in terms of accuracy. To address the issue of decreased accuracy for regular cameras, deep learning techniques (such as Convolutional Neural Networks, or CNNs) can be used to develop the algorithms that process the images captured by the cameras more accurately.

By using deep learning techniques, it is also possible to perform gaze tracking under challenging conditions, such as when low-resolution images (such as from webcams) are used, when there are variations in lighting, when the image backgrounds are varied, and when head movement is unrestrained.

In future, Sightcorp may experiment with using deep learning to enhance the gaze tracking component of DeepSight SDK.

Here are other articles you might find interesting:

 

   

Technical Specifications

The table below shows how FaceMatch SDK performs on the Labelled Faces in the Wild (LFW) dataset:

FPRTPRThreshold (Inverse of distance)
0.10.99900 ±0.002130.55448
0.010.99667 ±0.005370.59791
0.0010.99367 ±0.006050.62989

FPR = False Positive Rate
TPR = True Positive Rate

These results are an indication only and are based on the specific dataset Labelled Faces in the Wild. Customers can expect similar performance, with possible variations due to hardware and the availability of annotated data.