Loading…
Loading…
Follow individuals within every camera view, then across the site. Visylix Person Tracking uses appearance-based embeddings to hold a consistent identity through occlusion and crowding, and version 1 cross-camera appearance search assembles a person's journey across your other cameras as operator-reviewed candidates.
Visylix Person Tracking fuses real-time object detection with an appearance embedding model that encodes each person's visual appearance into a compact descriptor, maintaining a single continuous track for each individual within a camera view. Trajectory analysis maps movement paths through that view, revealing dwell zones, transition patterns, and flow bottlenecks. Cross-camera appearance search and person journeys ship today as a version 1 capability: from any detection, Visylix finds appearance matches on the site's other cameras within a time window and assembles a chronological path with travel times between cameras. Because it is appearance-based, it returns candidates for an operator to review rather than asserting identity automatically, and a purpose-trained re-identification model for higher precision is in our training pipeline. The model handles appearance changes caused by lighting shifts, partial occlusion, and crowded scenes in environments ranging from shopping malls and airports to corporate campuses and stadiums, and it runs entirely on-premise.
Engineered for accuracy, speed, and seamless integration into enterprise workflows.
Maintain a unified identity for each person throughout a camera view, recovering the track after occlusions and brief exits from frame.
An appearance embedding captures clothing color, texture, and body shape for robust matching under varying conditions.
Pick a person from any detection and find appearance matches on the site's other cameras within a time window. Results are candidates for operator review, not automatic identity assertions. A purpose-trained re-identification model for higher precision is in development.
Assemble a chronological, cross-camera path for a chosen person from those appearance matches, with travel times between cameras. Available in plain language through Radha ("where did this person go?").
Reconstruct full travel paths within a facility, identifying entry points, exit points, dwell times, and the sequence of zones each person visits.
Define custom zones on a floor plan and automatically record every transition between them, powering heatmap and flow-diagram visualizations.
Predictive motion models and appearance re-matching recover tracks after brief occlusions caused by pillars, crowds, or other obstructions.
Process up to 30 fps per camera with GPU-optimized inference, supporting simultaneous tracking across hundreds of concurrent streams.
Deployed across industries to deliver measurable value from day one.
Map shopper journeys from entrance to checkout. Understand aisle traffic, dwell times at displays, and the path-to-purchase for layout optimization and conversion improvement.
Track persons of interest as they move through a multi-camera network. Automatically generate alerts when subjects enter restricted zones or deviate from expected routes.
Monitor people flow through choke points, corridors, and gathering areas. Detect congestion before it becomes dangerous and guide real-time crowd control decisions.
Understand occupancy patterns floor by floor and room by room. Feed data into HVAC, lighting, and elevator systems to optimize energy usage and occupant comfort.
Combine multiple models for comprehensive video intelligence coverage.
Discover how Visylix AI models can transform your video infrastructure into an intelligent, automated system.