Tired of parking tickets draining your wallet? Meet PARKR, the ultimate hands-free parking defense system. Using a dashcam and real-time AI detection, PARKR keeps an eye out for parking enforcement vehicles, pays for the shortest required time, and notifies you before the ticket hits your windshield. Stop guessing when enforcement is near—let PARKR handle it. Smart, seamless, and always one step ahead.


Project Overview
Goal: Develop an AI-powered dashcam system that detects parking enforcement vehicles by recognizing their flashing lights and visit frequency. The system will automatically pay for parking when necessary and notify the user of detected enforcement officers.

Core Features
✔ Live Dashcam Video Processing – Detects enforcement vehicle lights in real time.
✔ AI-Based Light Recognition – Identifies flashing patterns to distinguish enforcement vehicles.
✔ Location Tracking & Visit Mapping – Logs enforcement presence over time.
✔ Automated Payment System – Pays for the shortest parking session upon detection.
✔ Instant User Alerts – Notifies the user via mobile app or text when enforcement is detected.

Technology Stack
Hardware Requirements
- Dashcam – Any high-resolution camera with night vision support ( I used the Raspberry Pi Camera v3)
- Edge Computing Device – Raspberry Pi 4B
- Wi-Fi / LTE Module – Enables live data transmission and app connectivity
- GPS Module – Logs enforcement locations for tracking visit frequency

Software & AI Model
- Operating System: Raspberry Pi OS
- Computer Vision Framework: OpenCV for image processing
- Object Detection Model: YOLOv8 (You Only Look Once) for real-time flashing light detection
- Machine Learning Framework: PyTorch for training the detection model
- Data Annotation Tool: LabelImg for creating training datasets
- Programming Languages: Python for AI & backend logic, JavaScript/React for the user app
- Database: Firebase for logging visit frequency
- Parking Payment API: Integration with municipal or third-party parking payment systems (e.g., PayByPhone, ParkMobile)

4. System Architecture
- Dashcam Video Feed → Captured in real-time.
- Image Processing → OpenCV processes frames to enhance visibility of flashing lights.
- Light Detection Model → YOLOv8 classifies detected flashing lights and determines enforcement presence.
- Visit Frequency Logger → Stores GPS-tagged enforcement visits for data insights.
- Automated Payment Trigger → Sends payment request via parking app API upon confirmed detection.
- User Notification System → Alerts the user via a mobile app, email, or SMS.


5. Development Phases
Phase 1: Research & Data Collection
Gather videos of parking enforcement vehicles at different times & weather conditions.
Annotate flashing lights & create training datasets.

Phase 2: AI Model Training
Train YOLOv8 on the annotated dataset to recognize flashing patterns.
Test and refine the model using real-world footage.

Phase 3: Software & Hardware Integration
Set up the Raspberry Pi to process dashcam footage.
Implement light detection and GPS logging functionality.
Develop backend logic to trigger payments.

Phase 4: Mobile App Development
Build a simple app for real-time alerts and enforcement activity logs.
Implement a notification system for manual double-checking.

Phase 5: Testing & Optimization
Test the system in various urban locations.
Optimize the detection model for low-light conditions.

Phase 6: Deployment & Scaling
Integrate with major parking apps.
Enhance system response time and add more enforcement recognition features.