DTK LPR SDK

Current Version: 6.0.2
Release Date: 5 May 2026

DTK LPR SDK (Automatic Number Plate Recognition / License Plate Recognition) is a comprehensive software development kit designed for developers who need to integrate reliable vehicle identification capabilities into their applications.

The SDK includes a high-performance LPR engine that detects and recognizes license plates from multiple input sources, including IP cameras, video files, capture devices, and still images (JPEG, PNG, BMP). Built using advanced, state-of-the-art algorithms, the engine delivers fast and stable operation with high recognition accuracy on real-time video streams up to Full HD (1920 × 1080) resolution. It supports a wide range of license plate formats from numerous countries.

In addition to license plate recognition, the SDK includes vehicle make and model recognition, providing extended vehicle metadata to support advanced analytics, monitoring, and intelligent traffic solutions.

It supports standard IP camera protocols and video encodings (e.g., MJPEG/HTTP, H.264/RTSP) as well as custom video sources, delivering decoded (RAW) frames to the LPR engine for processing. The SDK can be used with the most popular programming languages like C++, C#, Java, Python, VB.NET, Delphi, etc. The documentation, help files, and sample applications (C++, C#, Java, Python) are included in the SDK.

Vehicle Make & Model Recognition
The SDK with vehicle make and model recognition and vehicle view classification (front/rear), providing richer vehicle metadata alongside license plate results.
The make and model detection is performed in the region around the detected license plate, ensuring efficient processing and reliable association between the plate and the corresponding vehicle. This approach improves accuracy while maintaining real-time performance.
In addition, the SDK automatically determines whether the vehicle is captured from the front or rear view, which is useful for multi-camera tracking, traffic analysis, and entry/exit scenarios.

NVIDIA GPU / CUDA Acceleration Support
DTK LPR SDK includes optional support for NVIDIA GPU acceleration to significantly boost inference performance for high-throughput or real-time video processing workloads.

• For maximum performance, DTK LPR SDK uses NVIDIA TensorRT to optimize and run neural network inference on GPUs.
• Supported on systems with NVIDIA GPUs that support CUDA with a minimum Compute Capability of 7.5
• Requires installation of the appropriate CUDA Toolkit, TensortRT runtimes and NVIDIA drivers compatible with your GPU.
• On supported GPUs, the SDK uses GPU execution to accelerate neural network inference several times compared to CPU-only performance

DTK Software offers an end-user license plate recognition solution developed based on DTK LPR SDK, which can be integrated with existing systems without software development.

Create Free Trial Account

Online Demo

Main Advantages

• 98% recognition rate (including low quality and damaged plates)
• 15-40 ms processing of a single frame/image (on CPU)
• Recognition up to 16 plates on a single image
• Supports 2-line plates
• Detectd vehicle make/model and view (front/rear)
• Detects country format of a vehicle plate
• State detection for United States and Canada
• Considers country format specific features
• Provides coordinates and confidence of a recognized vehicle plate
• Provides coordinates and confidence of a individual characters of a vehicle plate
• Detects vehicle/plate movement direction in a video
• Captures video from IP cameras, video files, video capture devices

private void btnRecognizeVideo_Click(object sender, EventArgs e)
{
    LPRParams parameters = new LPRParams();
    parameters.Countries = "DE,NL,DK,PL";
    parameters.MinPlateWidth = 80;
    parameters.MaxPlateWidth = 300;
    LPREngine engine = new LPREngine(parameters, true, OnLicensePlateDetected);
    VideoCapture videoCap = new VideoCapture(OnFrameCaptured, OnCaptureError, engine);
    videoCap.StartCaptureFromIPCamera("rtsp://user:pass@192.168.1.105/axis-media/media.amp");
}
private void OnFrameCaptured(VideoCapture videoCap, VideoFrame frame, object customObject)
{
    LPREngine engine = (LPREngine)customObject;
    engine.PutFrame(frame, 0);
}
private void OnLicensePlateDetected(LPREngine engine, LicensePlate plate)
{
    Console.WriteLine(string.Format("Plate text: {0} Country: {1} Confidence: {2}",
           plate.Text, plate.CountryCode, plate.Confidence));
    plate.Dispose();
}

private void btnRecognizeImage_Click(object sender, EventArgs e)
{
    LPRParams parameters = new LPRParams();
    parameters.Countries = "DE,NL,DK,PL";
    parameters.MinPlateWidth = 80;
    parameters.MaxPlateWidth = 300;
    LPREngine engine = new LPREngine(parameters, false, null);
    List<LicensePlate> plates = engine.ReadFromFile("C:/Images/test.jpg");
    foreach (LicensePlate plate in plates)
    {
        Console.WriteLine(string.Format("Plate text: {0} Country: {1} Confidence: {2}",
            plate.Text, plate.CountryCode, plate.Confidence));
    }
}

Supported countries

Links

• Supported countries/formats (PDF)
• SDK Documentation
• Тo read about how to properly install the LPR camera, click on this link.

Supported Operating Systems

• Windows 64-bit (7/8/10/11/Server 2008/2012/2016/2019/2022)
• Linux (64bit, armhf, arm64)

Downloads

To download the full or trial version of the software, please go to the Downloads page.