azure-cognitive-csharp

Azure based Computer Vision

Object detection in camera stream with Azure Custom Vision service.

Prerequisites

1. Azure Subscription

First of all you need active Azure Subscription. In case you don't have one yet just follow this instruction: Create your Azure free account today

2. Azure Custom Vision

The solution based on Create your Azure free account today. You need to create custom vision resource to run the example:

3. Nuget packages

The solution uses the following nuget packages:

• Microsoft.Azure.CognitiveServices.Vision.ComputerVision
• Newtonsoft.Json
• OpenCvSharp3

Run the sample

Clone the repository:

git clone https://github.com/vzhukov/azure-bootcamp-2019.git

Open solution azure-cognitive-csharp\AzureLiveVideoStreamAnalysis.sln in Visual Studio.

Restore nuget-packages:

Open App.config and edit update app setting AzureCognitiveEndpoint with your value from Azure Custom Vision service:

Update app settings AzureCognitiveKey with your key:

Update app settings CameraIndex with the device index wich you will use (default web camera index - 0).

Start the application to detect objects:

How it works

Init video capture for default device (index = 0)

car reader = new VideoCapture(0);

Init Azure custom vision client

var visionClient = new ComputerVisionClient("AzureCognitiveKey")
    {
        Endpoint = "AzureCognitiveEndpoint"
    };

Read frame from camera:

var image = new Mat();
reader.Read(image);

Send image to Azure to recognize objects:

var imageBytes = image.ToBytes(".jpg");
using (var ms = new MemoryStream(imageBytes))
{
    var task = await visionClient.DetectObjectsInStreamAsync(ms);
}

Parse the result:

var items = task.Objects;
// item.ObjectProperty
// item.Rectangle