Passport: Edge AI Core Powering Cognitive Retail
Passport is our foundational Edge AI platform that performs real-time, large-scale computer vision and deep learning inference on in-store video streams from cameras or a store’s Video Management System (VMS), unlocking actionable intelligence from massive volumes of structured and unstructured data and seamlessly feeding insights into the entire MindGlyph solution suite.
By fusing deep learning, computer vision, and contextual spatial analytics, Passport delivers intelligence on shopper demographics, identity, behavior, shoplifting detection, vehicle fingerprinting, and more—routing processed data to Nexus for analytics and visualization, and to Dealz! and Scout for real-time engagement and alerts.
Passport Core Architecture Overview
Platform Capabilities
Real-Time Scalable Inference
Low-latency AI inference across hundreds of cameras.
Intelligent Video Analytics
Demographics, behaviors, detection, and profile analytics.
Real-Time Actionable Alerting
Loss prevention and clienteling event dispatch via Scout & Dealz!.
Flexible Deployment
Modular on-prem, VMS hosts, and hardware configurations.
Lambda Architecture
Dual-layer processing pairing real-time low latency with batch accuracy.
Scalable Data Aggregation
Local and cloud intelligence warehousing for downstream analysis.
Real-Time Scalable Inference
Performs scalable, low-latency AI inference on video streams across hundreds of in-store cameras or multiple VMS platforms, supporting on-prem, hybrid, or cloud configurations based on hardware capabilities.
Intelligent Video Analytics
Performs face recognition, faceless recognition, race, age, and gender identification, shoplifting detection, license plate recognition, vehicle fingerprinting, and shopper movement/dwell-time measurement.
Real-Time Actionable Alerting
Delivers real-time loss prevention and clienteling alerts through Scout, and sends personalized shopper engagement alerts through Dealz!.
Flexible Deployment
Supports ONVIF-compliant and non-compliant cameras, as well as a variety of VMS hosts, enabling seamless integration with existing store infrastructure. Its modular, on-premise deployment architecture allows AI processing to run on the cameras, the store’s VMS host, or MindGlyph Vista with our proprietary VMS, delivering low latency, high bandwidth, and significant cloud cost savings.
- ONVIF-compliant cameras with edge AI: partial or full deep learning processing runs on the camera, the store’s VMS, or MindGlyph Vista with its proprietary VMS.
- ONVIF-compliant cameras without edge AI: deep learning runs on store’s VMS or MindGlyph Vista.
- Non-ONVIF-compliant cameras with ONVIF VMS: deep learning runs on store’s VMS or MindGlyph Vista.
- Non-ONVIF cameras and non-ONVIF VMS: processing only possible if the VMS host has sufficient storage and compute.
Lambda Architecture
Combines batch and real-time data processing to efficiently handle massive volumes of in-store data, creating a scalable, fault-tolerant big data system that balances low latency for real-time insights with high accuracy from batch processing, powering robust AI inference and analytics across the solution suite.
Scalable Data Aggregation
Stores processed intelligence locally or in the cloud for downstream consumption by Nexus, supporting advanced analytics, reporting, and visualization across stores.