Hardware-accelerated tracking, localized facial recognition, and continuous spatial intelligence — running entirely in your browser. Built to be the ultimate hands-free vision companion.
Three foundational layers work in parallel, each decoupled, each purpose-built for zero-lag local inference.
The core routing layer seamlessly swaps between two massive frameworks without dropping the video feed.
ort.env.wasm.numThreads = 1 · WebGL + WASMA customized mathematical engine that converts 2D pixels into physical reality using strict metric focal ratios.
FOCAL_LENGTH_PX = 700 · Metric CalibratedCross-cycle matrix tracking captures structural changes in the room and exports them to timestamped local files.
detection_YYYY-MM-DD.txt logs locallydetection_2026-06-18.txt · Auto-timestampedUnlike generic cloud scanners, the Face VIP Interceptor runs completely parallel to the main object scanner with strict CPU-throttling to prevent resource depletion.
When COCO-SSD detects a generic "person," the neural network intercepts the frame. It extracts feature vector encodings, cross-references them against localized vector folders, and instantly overwrites the generic label with the mapped identity.
@vladmandic/face-api · 1000ms CPU Cooldown · Parallel ThreadTwo distinct systems: a silent background perimeter guard, and manual high-fidelity cloud overrides for complex scene analysis.
Dedicated to physical safety. Targets high-mass moving hazards and responds with overlaid visual warnings and emergency audio alerts.
Manual, high-fidelity overrides for complex scene analysis and dense text extraction from the environment.
Details regarding the physical layout and monitoring setup required to run Virtual Eyes optimally.
Strictly confined to home use. Perimeter monitoring keeps you focused while OCR extracts textbook diagrams.
Local Study Companion · ActiveSeamlessly swap between your built-in webcam and external tracking lenses on the fly via the nav hardware selector.
Hot-Swap · No Feed Drop