public static Iteration CreateTrainedImageClassificationProject(ICustomVisionTrainingClient client, string predictionResourceId, Guid?domain = null)
{
#if RECORD_MODE
var projName = ProjectPrefix + Guid.NewGuid().ToString();
// Create a project
var projDomain = domain.HasValue ? domain : GeneralDomain;
var project = client.CreateProject(projName, null, projDomain);
// Create two tags
var tagOne = client.CreateTag(project.Id, "Tag1");
var tagTwo = client.CreateTag(project.Id, "Tag2");
// Add five images for tag 1
var imagePath = Path.Combine("TestImages", "tag1");
var imageFileEntries = new List <ImageFileCreateEntry>();
foreach (var fileName in Directory.EnumerateFiles(imagePath))
{
imageFileEntries.Add(new ImageFileCreateEntry(fileName, File.ReadAllBytes(fileName)));
}
client.CreateImagesFromFiles(project.Id, new ImageFileCreateBatch(imageFileEntries, new List <Guid>(new Guid[] { tagOne.Id })));
// Add five images for tag 2
imagePath = Path.Combine("TestImages", "tag2");
imageFileEntries = new List <ImageFileCreateEntry>();
foreach (var fileName in Directory.EnumerateFiles(imagePath))
{
imageFileEntries.Add(new ImageFileCreateEntry(fileName, File.ReadAllBytes(fileName)));
}
client.CreateImagesFromFiles(project.Id, new ImageFileCreateBatch(imageFileEntries, new List <Guid>(new Guid[] { tagTwo.Id })));
// Train
var iteration = client.TrainProject(project.Id);
while (iteration.Status != "Completed")
{
Thread.Sleep(1000);
iteration = client.GetIteration(project.Id, iteration.Id);
}
Assert.True(client.PublishIteration(project.Id, iteration.Id, ClassificationPublishName, predictionResourceId));
// Get latest iteration
iteration = client.GetIteration(project.Id, iteration.Id);
// Make one prediction
string imageUrl = "https://raw.githubusercontent.com/Microsoft/Cognitive-CustomVision-Windows/master/Samples/Images/Test/test_image.jpg";
client.QuickTestImageUrl(project.Id, new ImageUrl(imageUrl), iteration.Id);
// Flush and re-init so we don't get all of the test setup in the session.
HttpMockServer.Flush();
HttpMockServer.Initialize(HttpMockServer.CallerIdentity, HttpMockServer.TestIdentity, HttpRecorderMode.Record);
return(iteration);
#else
return(null);
#endif
}
public async void TrainAndPublishProject()
{
using (MockContext context = MockContext.Start(this.GetType()))
{
HttpMockServer.Initialize(this.GetType(), "TrainAndPublishProject", RecorderMode);
using (var project = CreateTrainedImageClassificationProject())
{
ICustomVisionTrainingClient client = BaseTests.GetTrainingClient();
// Remove the last trained iteration so we can retrain
var iterationToDelete = client.GetIteration(project.ProjectId, project.IterationId);
var originalPublishName = iterationToDelete.PublishName;
await client.UnpublishIterationAsync(project.ProjectId, iterationToDelete.Id);
await client.DeleteIterationAsync(project.ProjectId, iterationToDelete.Id);
// Need to ensure we wait 1 second between training calls from the previous project.Or
// We get 429s.
Thread.Sleep(1000);
var trainedIteration = await client.TrainProjectAsync(project.ProjectId);
Assert.NotEqual(iterationToDelete.Name, trainedIteration.Name);
Assert.NotEqual(iterationToDelete.Id, trainedIteration.Id);
Assert.NotEqual(Guid.Empty, trainedIteration.Id);
Assert.True("Staging" == trainedIteration.Status || "Training" == trainedIteration.Status);
Assert.False(trainedIteration.Exportable);
Assert.Null(trainedIteration.PublishName);
// Wait for training to complete.
while (trainedIteration.Status != "Completed")
{
Thread.Sleep(1000);
trainedIteration = client.GetIteration(project.ProjectId, trainedIteration.Id);
}
// Verify we can republish using same name
Assert.True(client.PublishIteration(project.ProjectId, trainedIteration.Id, originalPublishName, BaseTests.PredictionResourceId));
project.IterationId = trainedIteration.Id;
}
}
}
public bool Treinar(string idDoProjeto, IEnumerable <Tag> tags)
{
if (!PodeTreinarProjeto(tags))
{
return(false);
}
var treinamento = _servicoCognitivoDeVisaoPersonalizadaTreinamento.TrainProject(new Guid(idDoProjeto));
while ("Training".Equals(treinamento.Status))
{
Thread.Sleep(1000);
treinamento = _servicoCognitivoDeVisaoPersonalizadaTreinamento.GetIteration(new Guid(idDoProjeto), treinamento.Id);
}
EnviarResultadosParaPedicao(idDoProjeto, treinamento);
return(true);
}
public static Iteration CreateTrainedObjDetectionProject(ICustomVisionTrainingClient client, string predictionResourceId)
{
#if RECORD_MODE
Dictionary <string, double[]> fileToRegionMap = new Dictionary <string, double[]>()
{
// FileName, Left, Top, Width, Height
{ "fork_1", new double[] { 0.219362751, 0.141781077, 0.5919118, 0.6683006 } },
{ "fork_2", new double[] { 0.115196079, 0.341127485, 0.819852948, 0.222222224 } },
{ "fork_3", new double[] { 0.107843138, 0.128709182, 0.727941155, 0.71405226 } },
{ "fork_4", new double[] { 0.148284316, 0.318251669, 0.7879902, 0.3970588 } },
{ "fork_5", new double[] { 0.08210784, 0.07805559, 0.759803951, 0.593137264 } },
{ "fork_6", new double[] { 0.2977941, 0.220212445, 0.5355392, 0.6013072 } },
{ "fork_7", new double[] { 0.143382356, 0.346029431, 0.590686262, 0.256535947 } },
{ "fork_8", new double[] { 0.294117659, 0.216944471, 0.49142158, 0.5980392 } },
{ "fork_9", new double[] { 0.240196079, 0.1385131, 0.5955882, 0.643790841 } },
{ "fork_10", new double[] { 0.25, 0.149951011, 0.534313738, 0.642156839 } },
{ "fork_11", new double[] { 0.234068632, 0.445702642, 0.6127451, 0.344771236 } },
{ "fork_12", new double[] { 0.180147052, 0.239820287, 0.6887255, 0.235294119 } },
{ "fork_13", new double[] { 0.140931368, 0.480016381, 0.6838235, 0.240196079 } },
{ "fork_14", new double[] { 0.186274514, 0.0633497, 0.579656839, 0.8611111 } },
{ "fork_15", new double[] { 0.243872553, 0.212042511, 0.470588237, 0.6683006 } },
{ "fork_16", new double[] { 0.143382356, 0.218578458, 0.7977941, 0.295751631 } },
{ "fork_17", new double[] { 0.3345588, 0.07315363, 0.375, 0.9150327 } },
{ "fork_18", new double[] { 0.05759804, 0.0894935, 0.9007353, 0.3251634 } },
{ "fork_19", new double[] { 0.05269608, 0.282303959, 0.8088235, 0.452614367 } },
{ "fork_20", new double[] { 0.18259804, 0.2136765, 0.6335784, 0.643790841 } },
{ "scissors_1", new double[] { 0.169117644, 0.3378595, 0.780637264, 0.393790841 } },
{ "scissors_2", new double[] { 0.145833328, 0.06661768, 0.6838235, 0.8153595 } },
{ "scissors_3", new double[] { 0.3125, 0.09766343, 0.435049027, 0.71405226 } },
{ "scissors_4", new double[] { 0.432598025, 0.177728787, 0.18259804, 0.576797366 } },
{ "scissors_5", new double[] { 0.354166657, 0.210408524, 0.305147052, 0.625817 } },
{ "scissors_6", new double[] { 0.368872553, 0.234918326, 0.3394608, 0.5833333 } },
{ "scissors_7", new double[] { 0.4007353, 0.184264734, 0.2720588, 0.6862745 } },
{ "scissors_8", new double[] { 0.319852948, 0.0339379422, 0.455882341, 0.843137264 } },
{ "scissors_9", new double[] { 0.295343131, 0.259428144, 0.403186262, 0.421568632 } },
{ "scissors_10", new double[] { 0.341911763, 0.0894935, 0.351715684, 0.828431368 } },
{ "scissors_11", new double[] { 0.2720588, 0.131977156, 0.4987745, 0.6911765 } },
{ "scissors_12", new double[] { 0.186274514, 0.14504905, 0.7022059, 0.748366 } },
{ "scissors_13", new double[] { 0.05759804, 0.05027781, 0.75, 0.882352948 } },
{ "scissors_14", new double[] { 0.181372553, 0.112369314, 0.629901946, 0.71405226 } },
{ "scissors_15", new double[] { 0.256127447, 0.190800682, 0.441176474, 0.6862745 } },
{ "scissors_16", new double[] { 0.261029422, 0.153218985, 0.513480365, 0.6388889 } },
{ "scissors_17", new double[] { 0.113970585, 0.2643301, 0.6666667, 0.504901946 } },
{ "scissors_18", new double[] { 0.05514706, 0.159754932, 0.799019635, 0.730392158 } },
{ "scissors_19", new double[] { 0.204656869, 0.120539248, 0.5245098, 0.743464053 } },
{ "scissors_20", new double[] { 0.231617644, 0.08459154, 0.504901946, 0.8480392 } }
};
// Find the object detection domain
var domains = client.GetDomains();
var objDetectionDomain = domains.FirstOrDefault(d => d.Type == "ObjectDetection" && d.Name == "General");
Assert.NotNull(objDetectionDomain);
// Create a new project
var project = client.CreateProject(ObjDetectionProjectName, null, objDetectionDomain.Id);
// Create two tags
var forkTag = client.CreateTag(project.Id, "fork");
var scissorsTag = client.CreateTag(project.Id, "scissors");
// Add all images for fork
var imagePath = Path.Combine("TestImages", "fork");
var imageFileEntries = new List <ImageFileCreateEntry>();
foreach (var fileName in Directory.EnumerateFiles(imagePath))
{
var region = fileToRegionMap[Path.GetFileNameWithoutExtension(fileName)];
imageFileEntries.Add(new ImageFileCreateEntry(fileName, File.ReadAllBytes(fileName), null, new List <Region>(new Region[] { new Region(forkTag.Id, region[0], region[1], region[2], region[3]) })));
}
client.CreateImagesFromFiles(project.Id, new ImageFileCreateBatch(imageFileEntries));
// Add all images for scissors
imagePath = Path.Combine("TestImages", "scissors");
imageFileEntries = new List <ImageFileCreateEntry>();
foreach (var fileName in Directory.EnumerateFiles(imagePath))
{
var region = fileToRegionMap[Path.GetFileNameWithoutExtension(fileName)];
imageFileEntries.Add(new ImageFileCreateEntry(fileName, File.ReadAllBytes(fileName), null, new List <Region>(new Region[] { new Region(scissorsTag.Id, region[0], region[1], region[2], region[3]) })));
}
client.CreateImagesFromFiles(project.Id, new ImageFileCreateBatch(imageFileEntries));
// Train
var iteration = client.TrainProject(project.Id);
while (iteration.Status != "Completed")
{
Thread.Sleep(1000);
iteration = client.GetIteration(project.Id, iteration.Id);
}
Assert.True(client.PublishIteration(project.Id, iteration.Id, ObjDetectionPublishName, predictionResourceId));
// Get latest iteration
iteration = client.GetIteration(project.Id, iteration.Id);
// Flush and re-init so we don't get all of the test setup in the session.
HttpMockServer.Flush();
HttpMockServer.Initialize(HttpMockServer.CallerIdentity, HttpMockServer.TestIdentity, HttpRecorderMode.Record);
return(iteration);
#else
return(null);
#endif
}