private static void CreateProjectFromImageData()
{
var projectPath = System.IO.Path.GetFullPath(@"..\..\..\");
var cvDataModelJsonFilePath = Path.Combine(projectPath, "Resources", "cv_data_model.json");
if (!File.Exists(cvDataModelJsonFilePath))
{
Console.WriteLine($"\tModel file was not found at: {cvDataModelJsonFilePath}");
Console.WriteLine("\tYou probably need to export the model first.");
return;
}
var trainingApi = new CustomVisionTrainingClient()
{
ApiKey = ConfigurationManager.AppSettings["Key"],
Endpoint = ConfigurationManager.AppSettings["Endpoint"]
};
// Read model from Local JsonFile
Console.WriteLine($"*** Loading local custom vision model from: {cvDataModelJsonFilePath} ***");
CustomVisionDataModel cvDataModel = null;
using (StreamReader r = new StreamReader(cvDataModelJsonFilePath))
{
var json = r.ReadToEnd();
cvDataModel = JsonConvert.DeserializeObject <CustomVisionDataModel>(json);
}
Console.WriteLine($"Tags to create: {cvDataModel.Tags.Count}");
Console.WriteLine($"Image to tag: {cvDataModel.TaggedImages.Count}");
Console.WriteLine("** Starting Setup of project **");
// Find the object detection domain
var domains = trainingApi.GetDomains();
var objDetectionDomain = domains.FirstOrDefault(d => d.Type == "ObjectDetection" && d.Name == "General (compact)");
// Create a new project
Console.WriteLine($"\tCreating new object detection project: {ConfigurationManager.AppSettings["ProjectName"]}");
var project = trainingApi.CreateProject(ConfigurationManager.AppSettings["ProjectName"], null, objDetectionDomain.Id, null);
// <snippet_tags>
Console.WriteLine("\tCreating tags");
// Create all the tags in the new project from the model
var tagsList = new Dictionary <string, Tag>();
cvDataModel.Tags.ForEach(tagLabel =>
{
var tag = trainingApi.CreateTag(project.Id, tagLabel);
tagsList.Add(tagLabel, tag);
});
// </snippet_tags>
Console.WriteLine("\tReading and Uploading images");
var index = 1;
cvDataModel.TaggedImages.ForEach(taggedImage =>
{
var fullPath = Path.Combine(projectPath, "Resources", "TrainingImages", taggedImage.FileName);
try
{
using (StreamReader r = new StreamReader(fullPath))
{
Console.WriteLine($"\tUploading image {index} of {cvDataModel.TaggedImages.Count}");
var image = trainingApi.CreateImagesFromData(project.Id, r.BaseStream).Images[0];
trainingApi.CreateImageRegions(project.Id, new ImageRegionCreateBatch()
{
Regions = taggedImage.Regions.Select(region =>
{
var tag = tagsList.FirstOrDefault(tagObject => string.Equals(tagObject.Key, region.Tag, StringComparison.OrdinalIgnoreCase));
return(new ImageRegionCreateEntry
{
TagId = tag.Value.Id,
ImageId = image.Image.Id,
Height = region.Height,
Left = region.Left,
Top = region.Top,
Width = region.Width
});
}).ToList()
});
}
}
catch (Exception)
{
Console.Error.WriteLine($"Error uploading image: {fullPath}");
}
index++;
});
Console.WriteLine("\tImages upload is done.");
// Now there are images with tags start training the project
Training(trainingApi, project, cvDataModel);
}
private static void Training(CustomVisionTrainingClient trainingApi, Project project, CustomVisionDataModel cvDataModel)
{
Console.WriteLine("\tTraining");
var iteration = trainingApi.TrainProject(project.Id);
// The returned iteration will be in progress, and can be queried periodically to see when it has completed
while (iteration.Status == "Training")
{
Thread.Sleep(1000);
// Re-query the iteration to get its updated status
iteration = trainingApi.GetIteration(project.Id, iteration.Id);
}
// The iteration is now trained. Publish it to the prediction end point.
var publishedModelName = "IotEdgeModel";
trainingApi.PublishIteration(project.Id, iteration.Id, publishedModelName, ConfigurationManager.AppSettings["ResourceId"]);
Console.WriteLine("\tDone training!");
}
