[00:00] Introduction

Vibecheck NYC

This is a real-time urban sensing system built with Google Cloud AI tools. It monitors public space across New York City using traffic camera feeds and generates a live surface of ambient pressure conditions.

Vibecheck reads the emotional temperature of the city, block by block, in motion.

Live NYC Emotional Temperature Map

Click any zone to select it for detailed analysis below. Hover for quick preview.

[00:30] Technical Architecture

Voronoi Zones & AI Processing Pipeline

Each zone is generated from a Voronoi map based on camera placement and borough geometry. Images stream through a proxy server and are processed using Cloud Vision.

Gemini then interprets each frame using a configuration of ambient tension indicators.

Zone Explorer - Select any zone above to analyze

Select a zone from the map above to see detailed analysis

[01:00] Adaptive Monitoring & Dashboard

Real-Time Zone Analysis

The system determines adaptive monitoring frequency. It increases resolution in areas of instability and zooms into quiet pockets inside tense zones.

These scores are logged in BigQuery for longitudinal modeling and visualized through Firebase.

Current Frame Analysis

Select a zone to view live feed
Monitoring Frequency: --

Emotional Temperature

--
Ambient Pressure

Zone: --

Borough: --

Cameras: --

[01:30] Agentic Evaluation Layer

Raw Vision Feedback & Temperature Calculation

Vibecheck doesn't just observe. It evaluates. The agentic layer monitors signal persistence and transition. It watches when patterns hold, when they collapse, and what may be reinforcing them.

Ambient Tension Indicators

Pedestrian Density

--

Number of people per frame area

Movement Velocity

--

Average speed of detected motion

Spatial Clustering

--

Degree of crowd formation

Obstruction Events

--

Blocked pathways or bottlenecks

Temperature Score Formula

$$T = \alpha \cdot \frac{D}{A} + \beta \cdot V^{-1} + \gamma \cdot C + \delta \cdot O$$

Where:
  • T = Emotional Temperature Score
  • D/A = Pedestrian Density (people per area)
  • V = Inverse Movement Velocity (higher = slower movement)
  • C = Spatial Clustering coefficient
  • O = Obstruction Events count
  • α, β, γ, δ = Weighted coefficients determined by ML training
[02:30] Data Storage & Classification

BigQuery Longitudinal Modeling

The same principles from neural circuit mapping can be applied to public infrastructure. Cities operate through feedback. The visible patterns are surface-level expressions of deeper circuits.

Zone Classification Pipeline

Current Zone Data Structure:

{
  "zone_id": "MN_001",
  "timestamp": "2024-01-15T14:30:00Z",
  "temperature_score": 7.2,
  "classification": "warm_tension",
  "indicators": {
    "pedestrian_density": 23.4,
    "movement_velocity": 1.8,
    "spatial_clustering": 0.65,
    "obstruction_events": 2
  },
  "adaptive_frequency": "15_minutes",
  "pattern_persistence": 0.8,
  "transition_probability": 0.23
}
                    

Classification Logic:

Cool Zones
T ≤ 3.0
Frequency: 1 hour
Warm Zones
3.0 < T ≤ 6.0
Frequency: 30 min
Hot Zones
T > 6.0
Frequency: 15 min
[03:00] Internal Monitoring & Agentic Layer

Real Gemini System Analysis & Database Viewers

The internal monitoring layer allows Gemini to audit its own functions, flag dropped inputs or unexpected stasis, and issue self-diagnostics through the dashboard.

This gives the system an interpretive voice that informs the people using it.

🤖 Gemini System Analysis

Gemini analyzes the system state with database access:

Gemini: Connected to Vibecheck system. I have access to Firebase collections, camera data, and system architecture. Ask me anything about the system state.

Live System Status:

Firebase Functions
Connecting...
Camera Database
Connecting...
BigQuery ML
Connecting...
Gemini AI
Connecting...

Live Database Viewers

Select a database viewer above to see live production data

[03:30] Real-Time Processing Tracker

Live Cloud Vision Processing Pipeline

This section shows the live processing pipeline for Cloud Vision API calls, input/output data, and system performance metrics.

🔬 Live Cloud Vision Processing Pipeline

🔄 Currently Processing

No active processing

⏳ Upcoming Zones

Loading queue...

📊 Recent Analyses - Raw Cloud Vision Output

Loading recent analyses...

🔄 Processing Pipeline Status

1
Fetch Image
Ready
2
Cloud Vision
Ready
3
Extract Objects
Ready
4
Calculate Score
Ready
5
Store Results
Ready
6
Update Zone
Ready

📈 System Performance

--
Total Processed
--
Success Rate
--
Avg Time (ms)
--
Active Cameras
[04:30] Cool Zones & Future Integration

Finding the Cool Zones

What Vibecheck is really searching for are the cool zones. Places where something about the physical, spatial, or social configuration produces atmospheric relief.

This work has a place in future versions of Google Maps. These factors influence how a walk feels, not just how fast it is.

System Architecture Overview

Built with Google Cloud AI Tools:

Gemini AI
Frame interpretation & system monitoring
Cloud Vision
Image processing & feature detection
BigQuery
Longitudinal modeling & classification
Firebase
Real-time dashboard & data sync

All of this was built in just a few days using the tools you've created.

Vibecheck is a case study in building agentic AI systems for civic-scale perception.

[Navigation] Explore System Components

🔗 Additional System Interfaces

Explore the complete Vibecheck NYC system through these specialized interfaces. Each page provides different views into the live production data and system performance.

🔬 Live Processing Tracker

Real-time Cloud Vision API monitoring with live pipeline visualization.

Features:
  • ✅ Live Vision API testing
  • 📊 Real-time pipeline monitoring
  • 🎯 Manual camera processing
  • 📈 Performance metrics
Open Live Tracker

🗄️ Database Viewers

Direct access to live Firestore and BigQuery data with real-time queries.

Open Database Views

🗺️ Geographic Visualizations

Interactive NYC maps with camera zones and real-time data overlays.

Explore Maps

📊 Analytics Dashboards

Comprehensive system dashboards with live metrics and analysis tools.

View Dashboards

🛠️ System Administration

Advanced system management and camera database tools.

Access Tools

⚡ System Status

Real-time system health monitoring and API status.

Current Status:
  • ✅ Cloud Vision API: Operational
  • ✅ Firebase Functions: Active
  • ✅ NYC Camera API: Connected
  • ✅ Database: 918 cameras live

💡 Quick Access: All interfaces connect to the same live production system. Data is synchronized across all views in real-time.

API Endpoints: Functions Health | System Status