WINK Streaming Releases Open-Source License Plate Recognition Model for Costa Rica

MIAMI, FL - April 17, 2026 - WINK Streaming has released WINK LPR, an open-source license plate recognition model trained specifically for Costa Rican license plates. The model achieves 98.4% plate-level accuracy and is available for free under a Creative Commons (CC BY 4.0) license.

            Key Highlights
            98.4% plate-level accuracy on Costa Rican license plates
4.4MB ONNX model — small enough for edge deployment
~5ms inference on CPU, ~2ms on GPU
Day and night camera support — color, IR, and grayscale feeds
CC BY 4.0 license — free for commercial and non-commercial use

        

Built for Real-Time Deployment

Built on a CTC architecture with CNN and BiLSTM layers, the 4.4MB ONNX model runs in approximately 5ms on CPU and 2ms on GPU, making it suitable for real-time deployment in traffic monitoring and smart city camera systems. The model handles both daytime color and nighttime IR/grayscale camera feeds, eliminating the need for separate models across lighting conditions.

Model Specifications

Architecture CTC with CNN + BiLSTM layers

Model Size 4.4MB ONNX

CPU Inference ~5ms per plate

GPU Inference ~2ms per plate

Plate-Level Accuracy 98.4%

Input Modalities Color, IR, grayscale

Available on Three Platforms

WINK LPR is available immediately on the major model-sharing platforms:

Hugging Face: xmichael/lpr-ocr-costa-rica
Kaggle: mikeixiixia/lpr-ocr-costa-rica

Part of the WINK Traffic & LPR Platform

WINK LPR is part of the company's broader WINK Traffic & LPR platform, which provides vehicle counting, wrong-way driver detection, emergency vehicle alerts, parked vehicle detection, and license plate reading on existing surveillance cameras for DOTs, public safety agencies, and smart city deployments. The open-source LPR model can be used standalone or integrated with WINK's commercial traffic analytics products.

Use Cases

Municipal traffic enforcement: Automated plate reading for tolling, parking, and citation systems
Smart city deployments: Vehicle access control and movement analytics
Public safety: Stolen vehicle detection and BOLO alerts on traffic camera networks
Research and academia: Foundation for further regional plate recognition research
Integration partners: VMS and analytics vendors building Costa Rican-specific solutions

License and Availability

WINK LPR is released under the Creative Commons Attribution 4.0 International license (CC BY 4.0), permitting both commercial and non-commercial use with attribution.

Documentation, weights, and inference code are bundled in each release.

About WINK Streaming

WINK Streaming builds secure cloud video infrastructure for government agencies, departments of transportation, law enforcement, and public safety organizations. The company's platform provides enterprise-grade video sharing, transcoding, and analytics with 24/7 reliability for mission-critical applications.

Contact Information

Website: https://wink.co
Contact: https://www.wink.co/contact-us
GitHub: github.com/winkmichael

WINK LPR for Costa Rica is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0).