Problem
Crowd build-up at stadium gates can turn dangerous quickly, and staff often react only once a zone is already overflowing. Operators need to see congestion forming early enough to act.
Computer Vision / Operations / Safety
Stadium - Crowd & Gate Monitoring with Computer Vision
See the crowd forming before it becomes a crush.
A computer-vision system that watches stadium gate zones in real time, flags crowding before it becomes dangerous, and recommends how to redistribute staff.
Evidence
See the code and verify the work yourself — nothing here is a claim you have to take on faith.
Solution
Stadium uses computer vision to count and track people across gate zones, classify each gate's status, and recommend where to move staff before a zone overflows.
My Role
Solo Developer
Impact
Stadium shows how computer vision on existing cameras can turn a safety blind spot into an early-warning system operators can act on.
My role
Solo Developer
Responsibilities focused on shaping the solution, connecting technical choices to user needs, and helping move the idea into a coherent working concept.
Built the full system end to end as a solo project
Implemented YOLO-based person detection and per-zone tracking
Designed the decision engine for gate status and alerts
Built the Flask API and the live monitoring dashboard
Technical architecture
How the solution was structured
Each case study is grounded in a practical technical approach, from local AI knowledge design to cloud-native analysis and behavioral analytics.
Source
Camera feed
01Processing
YOLO detection
02Intelligence
Zone & status logic
03Experience
Live dashboard & alerts
04Engineering notes
YOLO (Ultralytics) for person detection and tracking
Frame-by-frame assignment of people to configurable gate zones
A decision engine that classifies gate status and raises alerts
Flask API exposing live status at /api/status
HTML/JS dashboard polling the API every two seconds
Decisions
The calls that shaped it
The choices that mattered most — and the thinking behind each one.
01
Vision over new hardware
Used computer vision on ordinary camera feeds instead of installing new sensors — cheaper to deploy and it works with the cameras a venue already has.
02
Recommend an action, not just an alarm
The decision engine doesn't only flag a busy gate — it recommends where to move staff, so the output is something an operator can act on immediately.
Key features
What the project enables
Real-time person detection across four gate zones
Gate status: normal, busy, critical, overflow
Crowding alerts and incident logging
Staff distribution recommendations
ETA estimation
Live dashboard polling every two seconds
Impact
Applied value
Stadium shows how computer vision on existing cameras can turn a safety blind spot into an early-warning system operators can act on.
Inside the product
A look at how it works
A handcrafted preview of the experience — drawn to show the idea, not a stock screenshot.
StadiumGate monitor
Real-time
~2s dashboard refresh
4 zones
Configurable gate areas
Solo build
End to end by Abdulelah
Related projects
Continue exploring applied AI work
Move through the portfolio by domain, role, and technical approach.
Education AI / Local AI / NLP / LLMs
ChatUB - Local AI Academic Assistant
A local AI academic assistant for University of Bisha students, designed to answer questions based on official academic regulations, procedures, and documents.
View case studySustainability / Google Cloud / AI for Good
Althil - Urban Thermal Comfort Decision-Support Platform
A decision-support platform for improving urban thermal comfort by helping planners identify effective shade canopy locations based on sun paths, heat exposure, and location data.
View case study