iTrace

iTrace is an open-source Python toolkit for analysing gaze and pupil data, with optional webcam capture and browser-based local interfaces. It suits researchers and developers who need inspectable algorithms and can separate exploratory measurement from validated eye-tracking. Its strongest case is as a reference implementation and analysis workbench, not as a replacement for dedicated hardware.

Overview

The NumPy and SciPy core accepts gaze CSV files through a command-line workflow, calculates gaze in degrees of visual angle, and identifies fixations, saccades, and microsaccades. Capture is optional, so users with existing recordings can use the scientific layer without adopting the webcam components. Local FastAPI/WebSocket and Streamlit interfaces add browser viewing and dashboard-based inspection.

Its central limitation is measurement validation. The project reports algorithmic checks against synthetic ground truth, but not validation against real eye-tracking data or a reference tracker. It is consequently appropriate for method development and controlled comparisons; claims of device-level accuracy require comparison with a reference device or public dataset.

Key features

Gaze events and inspection

I-VT and I-DT methods detect fixations and saccades, while Engbert–Kliegl handles microsaccades. Visualisations include gaze traces, x/y time series, speed plots, event rasters, scanpaths, direction polar plots, amplitude histograms, and main-sequence scatter. That range lets researchers examine both raw trajectories and derived event structure rather than relying on a single summary.

Pupil processing

Tools cover blink detection and interpolation, preprocessing, and eye-crop segmentation. Size is reported in pixels or relative pupil-to-iris terms, which is useful for within-session or comparative analysis. Webcam pupil output is only a relative proxy and does not infer diameter in millimetres, so it is unsuitable when absolute physiological measurement is required.

Capture, interfaces, and exports

Optional capture uses OpenCV and MediaPipe Face Mesh iris landmarks. A local FastAPI and WebSocket HTML app, plus a Streamlit dashboard using Plotly, provide live views. The live workflow exports derived gaze, pupil, and capture-record CSV files together with a JSON analysis report. By default it does not save raw eye video or persistent eye-crop images; live viewing may send a transient JPEG crop to the browser over a local WebSocket, a sensible privacy posture for local work.

Pricing

iTrace is free and released under the MIT license. The sole plan, iTrace (MIT-licensed open-source software), costs free and is delivered as a Python package. There are no paid tiers, seat charges, quotas, renewal terms, or trial expiry to weigh. The trade-off is that users take on their own installation and validation work. Python installation is documented through uv, with capture, dashboard, figures, and web components supplied as optional extras, so a minimal analysis setup need not include the larger interface stack.

Platforms

The package targets Linux and self-hosted use, with a web browser available through the local HTML interface. Webcam input is the capture method. This is a practical fit for teams comfortable running Python services locally; it is not a hosted cross-device product with managed deployment.

Who it's for

Science and research users get transparent event-detection and pupil-processing methods, CSV/JSON interchange, and plots suited to algorithm comparison. Developers can use the separation between the NumPy/SciPy core and optional capture layer as a foundation for experiments. It is a poor choice for studies that need certified or reference-matched accuracy, millimetre pupil values, or turnkey participant operations without technical setup.

Pros and cons

Pros

• Open MIT licensing: the code and Python package can be adapted without subscription cost.
• Method breadth: I-VT, I-DT, Engbert–Kliegl, blink interpolation, and multiple diagnostic plots support serious exploratory analysis.
• Flexible data paths: analyse existing gaze CSV files or add optional webcam capture and local dashboards.
• Privacy-conscious defaults: derived exports omit raw eye video and stored eye-crop images.

Cons

• No real-device validation: synthetic and closed-loop checks do not establish performance against an eye tracker.
• Relative pupil proxy: webcam measurements cannot provide millimetre pupil diameter.
• Technical deployment: Linux, Python, optional extras, and self-hosted services suit technical users more than turnkey teams.
• Webcam dependence: capture quality and interpretation are bounded by a camera-based approach rather than dedicated hardware.

Alternatives

Choose GazePlotter instead when you want a free, GPLv3 open-source tool spanning Android, iOS, Linux, macOS, web, and Windows. Labvanced is the better fit for a hosted freemium research workflow with a free allowance of 300 MB, 10 recordings, and one active recruiting link, plus a trial. For non-commercial webcam use across Android, iOS, web, and Windows, consider GazeRecorder.

Hawkeye suits iOS and web users who prefer a managed freemium service (one project and five sessions on its free plan). Linux users seeking an open-source gaze tracker that requires a webcam and Linux build dependencies can choose OpenGazer. PyGaze is a free open-source Python toolbox for Linux, macOS, and Windows. Pick Chiasm when browser/API delivery and pay-per-use participant minutes are preferable. OGAMA is a free, open-source Windows alternative.

Verdict

iTrace is recommended for researchers and developers building or comparing eye-movement analysis methods who value open code, local operation, and broad event and pupil tooling. Its decisive advantage is a capable, modular scientific core at no cost. Look elsewhere when your project needs real-device validation, absolute pupil measurements, or a turnkey hosted workflow; iTrace itself makes those boundaries clear.