Eye Tracking Datasets for Gaze and Attention AI
Calibrated gaze annotation and saliency datasets across diverse participants, display types, and content categories. Built for gaze estimation, visual attention, and appearance-based eye tracking AI.
20+
Years in AI training data
1,000+
Language locales
1M+
Hours of video annotated
ISO
27001 certified
The Challenge
Beyond General Gaze Benchmarks
MPIIGaze and GazeCapture established webcam-based gaze estimation baselines. Production gaze and attention AI requires participant diversity, domain-specific content, and calibration protocols those benchmarks cannot provide.
Driver monitoring systems require gaze datasets across diverse driver demographics under real driving conditions. Clinical eye tracking AI requires patient populations with specific ocular conditions. Consumer attention AI requires real-world content viewing behavior.
Off-the-shelf gaze datasets suffer from participant demographic gaps (limited age, ethnicity, and ocular condition diversity) and content domain mismatches between benchmark stimuli and your application's content type.
LXT builds custom eye tracking datasets with your participant demographic targets, content stimuli, and capture protocol. We deliver calibrated gaze data and saliency labels that match your gaze AI application's requirements.
Why Teams Upgrade
Limitations of Public Eye Tracking AI Datasets
Standard benchmarks serve research well. Production deployments need more.
| Dataset | Primary Limitation | Impact |
|---|---|---|
| MPIIGaze | 15 participants only; office backgrounds; limited demographic diversity; appearance-based models overfit to this small sample | 15 participants |
| GazeCapture | 1,500 participants via crowdsourcing; variable device and environment quality; no calibrated gaze ground truth | Uncalibrated |
| Gaze360 | Outdoor scenes only; no screen-based content; limited applicability to product UI or document reading attention | Outdoor only |
| MIT Saliency | Computational saliency maps only; no ground truth eye tracking data; model-predicted rather than human-recorded gaze | No real gaze |
| SALICON | Mouse-click saliency approximations; not true gaze; restricted to MS-COCO images rather than domain-specific content | Mouse clicks |
Not sure which specs you need?
Our data specialists help you scope the right dataset for your model architecture.
Configurable Specifications
Specs Built Around Your Model
Public datasets come fixed. Yours is configured for your architecture, environment, and use case.
Participant Design
Gaze Diversity
- Demographics: Age, ethnicity, and corrective lens variation across participants
- Ocular Conditions: Normal vision, glasses, contacts, and clinical conditions per use case
- Sample Size: Minimum 50-500 participants depending on application requirements
Stimuli Content
Viewing Materials
- Domain Content: Your product interface, documents, videos, or scene types
- Controlled Design: Randomized presentation with fixation control and counterbalancing
- Free Viewing: Natural viewing behavior on target content without task constraints
Capture Configuration
Eye Tracking Setup
- Hardware: Remote eye tracker, wearable, or appearance-based webcam collection
- Calibration: 9-point and 16-point calibration protocols with drift correction
- Ground Truth: Fixation, saccade, and blink event classification from raw gaze streams
Need a custom configuration?
We've built datasets across dozens of domains and use cases. Let's scope yours.
Capturing Complexity
Edge Cases in Eye Tracking Datasets
High-accuracy models handle rare attributes that public datasets miss.
Glasses and Lens Reflections
Spectacle reflections create gaze estimation artifacts. Targeted participant collection with eyeglass wearers under varied lighting prevents deployment failures for glasses users.
Head Pose Variation
Gaze estimation accuracy degrades with head rotation. Collection across natural head pose ranges with ground truth 3D gaze vectors supports head-pose-robust model training.
Clinical Populations
Patients with nystagmus, strabismus, or other ocular conditions produce non-standard gaze patterns. Clinical collection with appropriate ethical protocols supports medical eye tracking AI.
Dynamic and Video Content
Video and animated content produce different gaze distributions than static images. Dynamic stimulus collection with frame-aligned gaze annotations supports video attention model training.
Ground Truth Quality
Human-in-the-Loop Annotation
Precise annotation bridges raw data and learnable signal. Expert annotators deliver precision automated tools can't match.
Calibrated Gaze Recording
Precise eye tracking data collected with validated hardware and calibration protocols. Raw gaze streams, fixation events, and saccade classifications delivered per participant.
Fixation and Saliency Annotation
Fixation density maps, attention heatmaps, and region-of-interest labels for saliency model training and UI attention analysis.
Participant Metadata
Demographic information, vision correction status, and session quality metrics included for participant-aware and demographic-controlled model training.
Industry Applications
Eye Tracking Datasets for Your Domain
Custom taxonomies and collection protocols for specific deployment contexts.
Driver Monitoring
Drowsiness, distraction, gaze direction monitoring
UI and UX Research
Interface attention, reading behavior, visual search
Clinical Eye Tracking
Neurological assessment, vision rehabilitation
Gaming and XR
Foveated rendering, gaze-based game control
Media Attention
Advertising attention, content effectiveness
Reading AI
Reading speed, comprehension, dyslexia screening
Assistive Technology
Gaze-based AAC communication devices
EdTech
Learning attention, student engagement monitoring
Compliance & Ethics
Secure and Ethical Data Collection
Data collection involving people and sensitive content requires robust security, compliance, and ethical protocols at every stage.
Global Demographic Reach
Collection across 1,000+ locales and diverse demographics to prevent algorithmic bias in your deployed models.
ISO 27001 Certified
Sensitive projects processed in certified secure facilities meeting the highest information security standards.
GDPR & Privacy Compliance
All collection and annotation protocols vetted for consent and privacy. Legally robust for global deployment.
Frequently Asked Questions
Eye Tracking Dataset FAQs
What eye tracking hardware do you use?+
We use Tobii, SR Research EyeLink, and GazePoint remote eye trackers for precision lab collection, and appearance-based webcam collection for large-scale datasets. Hardware choice is matched to your accuracy requirements.
Can you collect data on our specific content or interface?+
Yes. We design stimulus protocols around your product interface, documents, videos, or scene content. Stimulus design includes counterbalancing and fixation control for clean data collection.
How many participants do you typically need?+
For appearance-based gaze estimation training, 100-500 participants are typically needed. For saliency and attention studies, 30-100 participants per stimulus set produce stable fixation distributions.
Can you collect from clinical populations?+
Yes, with appropriate ethics approvals. We coordinate IRB or equivalent ethics review for clinical populations and work with clinical partners for specific ocular or neurological conditions.
What formats do you deliver gaze data in?+
ASC (EyeLink), TSV, and CSV formats with fixation, saccade, and blink event classifications. Scanpath data, attention heatmaps, and saliency maps delivered in standard image formats.
What does a custom eye tracking dataset cost?+
Projects range from $20K for focused webcam-based datasets (100-500 participants) to $100K+ for large precision eye tracker studies with clinical populations or complex stimulus design.
How do you handle participant privacy in gaze data?+
Participant identities are separated from gaze records. Video data (if captured) is held separately from gaze streams. All collection follows GDPR and applicable biometric data regulations.
Get Started
Scope Your Custom Eye Tracking Dataset
Share your participant requirements, content stimuli, and gaze accuracy needs. A data specialist will provide a detailed proposal within 48 hours.
Please provide us with the details of your inquiry and one of our team members will be in touch.
Join our global team of contributors today
Apply here to be considered for future projects including data collection, annotation and transcription
Start application