R&D testing · Open for research collaboration
CANINE BINOCULAR EYE TRACKING SYSTEM
Canine Binocular
Eye Tracking
Capture binocular eye dynamics to quantify visual attention and behavioral response — integrating binocular capture, environment view, head pose sensing, and AI analysis.
Lightweight wearTriple-view syncEdge AI analysis
WHY EYE TRACKING
From observing behavior
to understanding attention
Traditional canine research can tell what a dog did, but rarely where it looked, how long it attended, or when attention shifted during a task.
Synchronized eye, environment, and head-pose recording provides continuous, objective metrics for behavior research, working-dog evaluation, and human–dog interaction studies.
When does attention shift?
How do gaze and head motion relate?
SYSTEM ARCHITECTURE
A three-part complete system
From mobile capture to edge processing and AI analysis — a closed data loop.
SchematicLightweight head-mounted device
Integrates binocular eye cameras, forward environment camera, and head IMU for mobile capture.
- Total weight ≤ 200 g
- Silicone body with hidden cable routing
- Adjustable forward camera angle
SchematicEdge computing host
Local video ingest, encoding, streaming, storage, and AI inference — lower deployment friction.
- RK3588 · 6 TOPS
- Wi-Fi 6 high-speed transfer
- Triple-stream record, push & infer
SchematicBehaviorAtlas DogPupil Studio
Device connect, live preview, acquisition, recording management, pupil analysis, and visualization.
- Browser wireless live view
- Pupil diameter curves
- Marking records & event annotation
CORE CAPABILITIES
Core capabilities for real experiments
Stable capture first — turn hard-to-observe visual attention into analyzable data.
Triple-view synchronized capture
Synchronized left-eye, right-eye, and forward environment views at a stable 30 fps.
Wireless live view
Connect over Wi-Fi and preview feeds in the browser — fewer cable constraints.
Head-pose synchronization
Record 3D head angles and motion trends aligned with eye and environment data.
AI pupil detection
Automatically detect and track pupils for position, diameter, and trend analysis.
Reliable recording storage
Segmented recording with watchdog mechanisms for long-session reliability.
Analysis & visualization
Auto-generate pupil diameter curves and extend analysis dimensions as needed.
WORKFLOW
Two analysis paths for different experiments
Edge host streams, records, and runs AI inference in parallel — for live monitoring and instant feedback.
01Device fitting & adjustment→
02Triple-view live preview→
03Synced capture & inference→
04Realtime result display
PERFORMANCE
Validated key performance
Core hardware/software development is complete — now in multi-scenario optimization and field testing.
5-hour continuous recording stability testTriple USB camera synced record & streamThermal optimization under full load
APPLICATIONS
Application scenarios
A new observation dimension for canine cognition, training evaluation, human–dog interaction, and welfare research.
Canine cognition research
Analyze visual and pupil responses to people, objects, and varied stimuli.
↗Working-dog training evaluation
Assess attention state and stimulus response in police, search-and-rescue, and other working dogs.
↗Human–dog interaction research
Observe gaze changes during faces, gestures, eye contact, and command tasks.
↗Animal behavior & welfare
Combine gaze, head pose, and behavior data to study stress and environmental adaptation.
↗R&D PROGRESSFrom core development
From core development
to real-world validation
Core hardware and software iterations are complete — ongoing work focuses on wear stability, multi-breed fit, and pupil-model robustness.
Completed
✓ Core head-mount structural design
✓ Stable 30 fps triple-camera capture
✓ Low-latency wireless streaming
✓ BehaviorAtlas DogPupil Studio
✓ Pupil detection & curve visualization
In optimization
→ Multi-breed, age, and size fit
→ Wear stability during motion
→ Video stabilization & model robustness
→ AI model accuracy & speed
→ Battery monitoring & low-power alerts
OPEN FOR COLLABORATION
Open research collaboration & co-creation
We welcome teams in canine cognition, animal behavior, working-dog training, and human–dog interaction to validate capture across breeds, tasks, and environments.
Multi-breed testing · Paradigm design · Joint analysis