The Challenge
Parking violations were becoming a growing concern for the City of Ontario. Cars parked beyond allowed time limits in public lots and roadside spots were creating traffic management issues and lost revenue. Manual enforcement was costly, inefficient, and prone to human error.
Our Solution
Sketric Solutions designed PlateSniper - a computer vision–powered license plate recognition (LPR) system that runs in real time on edge devices such as Raspberry Pi 5, iPhones, and Android devices. The solution integrates seamlessly with both mobile and dash-mounted cameras. Mounted on taxis and Ubers, the system continuously collects city-wide data while vehicles are in motion, enabling scalable and efficient monitoring without reliance on cloud processing.
How It Works
Dash-mounted cameras on taxis/Ubers continuously scan for plates. Custom-trained YOLO models detect plates, and Parseq OCR extracts alphanumeric values. Each detection is timestamped, deduplicated, and stored in a Flask backend with database persistence.
Key Features & Capabilities
High Accuracy: 95%+ OCR accuracy, fine-tuned for Ontario plates
Real-Time Processing: 12 FPS on Raspberry Pi 5, 30 FPS on iOS with CoreML acceleration
Persistent Tracking: Custom ID tracking logic prevents duplicate entries
Scalable Deployment: Runs on both mobile (Flutter) and embedded (Rpi5) devices
User-Friendly App: Flutter app with optimized detection modes (tracking vs. interval)
Custom OCR Training: Parseq OCR trained on real-world noisy datasets for robustness
Tech Stack
Frontend: Flutter (mobile), React Native (web app/dashboard)
Backend: Flask + Python
Models: PyTorch, TensorFlow, Ultralytics YOLOv11
OCR: Parseq OCR (custom-trained)
Hardware: Raspberry Pi 5, iOS/Android native cameras
Networking: Dio HTTP client for fast uploads
Real-World Impact
Built an interactive dashboard in React Native + Web that allows enforcement teams to search, filter, and view all detected plates, track device IDs, timestamps, and locations, and view cropped license plate images for verification.
Key Highlights
Built a production-ready MVP that impressed stakeholders with demo results
Achieved real-time performance on RPi5, proving feasibility for large-scale enforcement
Designed a scalable architecture for future expansion into multiple North American cities
Client perspective“Sketric Solutions helped us turn a complex urban problem into a practical, AI-powered solution. Their expertise in computer vision and embedded AI proved invaluable.”
— PlateSniper Team
SiteTrax.io: OCR and Computer Vision for Intermodal Asset Visibility
How Sketric helped SiteTrax.io turn yard imagery into OCR-based asset records for intermodal visibility, yard checks, and logistics-system workflows.
Read case study →
Automated AR with OpenCV
Advanced augmented reality pipeline enabling seamless integration of virtual objects into real-world environments using computer vision.
Read case study →
Vergo AI Assistant
Vergo AI Assistant turns work video into pose landmarks, time-coded REBA signals, and a human review path for ergonomic screening without claiming medical diagnosis or validated accuracy.
Read case study →Have a project with similar engineering constraints?
Tell us what must be built, measured, integrated, and released. We’ll help define the technical path and the evidence needed to support production claims.