Open source gaze tracking

The short answer

Open source gaze tracking is software, with publicly available source code, that estimates where a person is looking from camera images of their eyes and face. You can run it with an ordinary webcam or pair it with dedicated eye-tracking hardware. The main limit is precision. Webcam-only tools generally estimate gaze less precisely than systems built around infrared illumination and dedicated eye cameras. They also depend more on lighting, head position and calibration. Gaze tracking works on video only. It does not capture or process speech, so it does not change which microphone your computer records from.

The evidence for this guide included no confirmed project names, version numbers, accuracy figures, prices or release dates. For that reason, it describes how these tools work and how to evaluate them, but it does not rank or compare specific projects. Check any figure you find in a project's own documentation or an independent evaluation before relying on it.

What "open source" means here

"Open source" describes the software license, not the quality or method of tracking. An open license lets you inspect, modify and usually redistribute the code. Licenses differ in what they allow. Some permit commercial use freely. Others require you to share any modifications, and some code is published under terms that restrict commercial use. Read the license file of any project before building it into a product or a workplace tool.

How the main approaches differ

Aspect | Webcam-based software | Dedicated eye-tracking hardware with open software

Camera | Standard visible-light webcam | Purpose-built eye cameras, often with infrared illumination

Typical precision | Coarser; suited to estimating a screen region | Finer; suited to more detailed gaze measurement

Sensitivity to conditions | Higher sensitivity to room lighting, glare, glasses and head movement | Generally more controlled, though fit and calibration still matter

Cost of entry | Uses equipment many people already own | Requires buying or building additional hardware

Typical uses | Experiments, prototypes, coarse gaze-aware interfaces | Research studies, finer interaction, detailed analysis

The table makes qualitative comparisons only. The supplied evidence included no measured accuracy figures, so it gives no numbers.

Use for accessibility and gaze interaction

Gaze interaction lets someone point, select or type by looking at elements on a screen. This can help people with limited hand or arm movement. Open source tools are useful for experimenting with this kind of interface and for adapting it to individual needs. When gaze is someone's primary means of communication or computer access, reliability matters more than in a prototype. Dedicated assistive systems, along with an assessment by a qualified clinician or assistive-technology specialist, may be more appropriate in that case. Open source webcam tools have not been shown to be dependable enough for that role in every setting, and the evidence for this guide does not establish that they are.

What to check before choosing a tool

• License: confirm it permits your intended use, especially commercial or redistributed use.
• Maintenance: check when the code was last updated and whether reported issues receive responses.
• Platform support: confirm the operating system, programming language and any browser requirements.
• Calibration: find out how calibration works, how long it takes and how often it needs repeating.
• Input hardware: confirm whether an ordinary webcam is enough or specific eye-tracking hardware is required.
• Processing location: check whether video is processed locally or sent elsewhere, since face and eye images are sensitive data.
• Output: confirm whether it provides screen coordinates, gaze direction, fixation events or only raw estimates, and whether that matches your application.
• Documentation: look for setup instructions and any published accuracy evaluation, and note the conditions under which accuracy was measured.

Practical conditions that affect results

Webcam-based gaze estimation depends on a clear view of the eyes. Even, front-facing light and minimal glare on glasses help. So does a camera positioned close to eye level, a stable head position and recalibration after you move the camera or change seating position. These are general principles rather than measured results for any particular tool.

What this means for capturing speech

Gaze tracking reads the webcam's video stream. A webcam's built-in microphone is usually exposed to the computer as a separate audio input device. Running a gaze tracker therefore does not select, mute or process your microphone. You still choose the audio input in your operating system or conferencing app as usual.

• Camera access conflicts: some operating systems and camera drivers let only one application use a camera at a time. A gaze tracker and a conferencing app may then compete for the same webcam, while the microphone keeps working independently.
• Processor load: video-based tracking uses computing resources. On a heavily loaded machine this could affect other real-time tasks, including calls. No measurements were supplied, so treat this as a possibility to watch for rather than a confirmed effect.
• Combined gaze and voice input: some interfaces use gaze to point and a spoken command to confirm. In that design, choosing a clear audio input matters as much as gaze calibration, and the two should be tested together.

Gaps in the available evidence

The material behind this guide contained no confirmed details about specific open source gaze tracking projects. It had no accuracy or latency measurements, hardware specifications, supported-platform lists, license terms or dates. It also contained no tests of gaze tracking running alongside webcam audio capture. For any of these points, consult the project's own documentation and published evaluations, and test with your own camera, lighting and conferencing setup.

Next step

Decide whether you need coarse, screen-region gaze estimates or finer measurement. If coarse estimates are enough, try a maintained webcam-based project with a license that fits your use, calibrate it under your normal lighting, and confirm that your conferencing app can still access the camera and your chosen microphone at the same time. If you need finer precision or dependable accessibility use, evaluate dedicated eye-tracking hardware with open software, or a purpose-built assistive system.