EventScope: Edge-to-Cloud Analytics for Live Event Intelligence

The Challenge

Johnson & Johnson’s Medical Division needed a way to understand how visitors engaged with their booths at a major international medical congress. Despite extensive investment in booth design and interactive content, organizers lacked quantitative insights into attendee behavior — such as dwell time, engagement with specific displays, and overall traffic flow.

Traditional methods like badge scans or manual counts provided only partial data, missing real behavioral patterns. The challenge was to deliver real-time, privacy-compliant analytics capable of operating in offline, high-traffic environments without relying on cloud connectivity.

The Solution

Infinity Loop developed EventScope, a distributed edge-to-cloud analytics platform purpose-built for live events. The system captures visitor interactions using computer vision and edge computing, transforming raw behavioral signals into actionable insights — all while maintaining strict GDPR compliance through on-device privacy safeguards.

EventScope empowers event teams to visualize engagement patterns, dwell times, and traffic density in real time, helping optimize staffing, content placement, and overall booth performance during the event.

Key Deliverables

Edge-to-Cloud Analytics Platform

A resilient, hybrid system that processes over 10,000 events per day with sub-100ms latency, enabling real-time insight generation.

Privacy-First Edge AI

All visual analysis occurs on-device with automatic face blurring and anonymous session tracking, ensuring full compliance with European data regulations.

Real-Time Dashboard

Built using Hotwire and Rails 8 Solid Stack, the dashboard provides live metrics, heatmaps, and engagement analytics across multiple booths and displays.

Zero-Touch Deployment

Ansible automation enables new devices to be deployed in under 20 minutes, reducing setup complexity and eliminating on-site IT overhead.

Technical Approach

EventScope runs entirely on IoT hardware, combining a lightweight Python + TensorFlow computer vision pipeline with a Rails 8 backend. Each device captures and processes video locally, extracts anonymised event data, and synchronises batches securely over a WireGuard VPN mesh.

The system’s offline-first architecture ensures reliability in venues with limited connectivity. All ML inference, event buffering, and synchronisation occur at the edge, while aggregated analytics and dashboards run on the central node.

This architecture balances performance, cost efficiency, and privacy, proving that high-value insights don’t require expensive cloud infrastructure.

Results

EventScope provided Johnson & Johnson’s event team with unprecedented visibility into booth engagement and visitor flow:

• 73-second average dwell time across exhibits
  
• Real-time performance metrics for 20+ locations
  
• Zero data loss during intermittent network conditions
  

The system provides real-time insights including:

• Total Visitors vs. Engaged Sessions — tracks overall footfall and identifies meaningful interactions.
  
• Average Dwell Time — measures how long attendees spend at each booth or display.
  
• Engagement Rate — calculates active participation versus passive browsing.
  
• Sentiment Analysis — detects audience mood and response through on-device emotion recognition.
  
• Interaction Counter — records gestures, touchpoints, and display interactions for behavioral mapping.
  

Beyond analytics, the platform demonstrated that edge-first systems can deliver enterprise-grade intelligence while maintaining both cost efficiency and data privacy — a critical requirement for medical events in regulated environments.

Technologies: Rails 8, Ruby 3.4, Hotwire, Tailwind, Python 3.11, FastAPI, TensorFlow Lite, MediaPipe, PostgreSQL, WireGuard, Ansible