SACASE
06Smart Parking
2026CLOSE CASE ×
06 / 07Mobility / Computer vision2026

Smart Parking System

A smart gate is useless if one weak camera read can trap the entire operation.

ROLE
System design · Full-stack engineering · Computer vision
STATUS
Private system build
STACK
React · Express · FastAPI · YOLOv8 · Computer vision
SYSTEM ARCHITECTURE / 06SYSTEM LOGIC / PRIVATE BUILD
ARRIVAL / MA-2458READ CONFIDENCE / REVIEW
MANUAL CONFIRMATIONOPERATOR OWNS THE UNCERTAIN STATEGATE / HOLD → RESOLVE → LOG
01 / CONTEXT

WHY THIS PRODUCT
NEEDED TO EXIST.

CHALLENGE

Camera reads, reservations, and real-world arrival data do not always agree. The system still needs a fast, human-confirmable path that protects both access and an auditable occupancy record.

RESPONSE

I unified plate detection, reservation lookup, QR-linked receipts, allocation, and operator confirmation into one event flow, with manual verification treated as a first-class state.

02 / OPERATING LOGIC

THE OPERATING
SEQUENCE.

  1. 01DETECT

    Read the arrival

    Camera, reservation, or QR data starts the event.

  2. 02VALIDATE

    Check access

    The service compares the incoming signal with reservation and occupancy state.

  3. 03CONFIRM

    Resolve uncertainty

    Operators can correct a plate or approve the valid path manually.

  4. 04LOG

    Close the event

    The gate result and occupancy change become auditable system state.

03 / PRODUCT DECISIONS

DECISIONS THAT
SHAPED THE BUILD.

01 / INPUT

Multiple arrival signals

A plate, reservation, or QR receipt can begin the same controlled entry flow.

02 / FALLBACK

Humans can resolve ambiguity

Weak recognition moves to an operator-confirmable state instead of silently denying access.

03 / RECORD

Every outcome closes

Entry, rejection, correction, and allocation become explicit events the operator can inspect.

04 / EVIDENCE LIMIT

THE SYSTEM.
BEYOND THE INTERFACE.

This case documents architecture, failure handling, and implementation decisions while the product interface and source remain private.

05 / MY RESPONSIBILITY

SCOPE OF
RESPONSIBILITY.

I designed the service boundaries across the web interface, Express/FastAPI services, and the YOLOv8-assisted plate pipeline.

The private build documents the operating logic and fallback path. Public interface evidence and a repository are not available, so this case study deliberately avoids invented screenshots.

  • Gate operations
  • Plate recognition
  • Manual fallback
  • Occupancy
NEXT CASE / 07Lead Automation