private CustomVisionResponse AddCustomVisionResponseToImage(CustomVisionResponse customVisionResponse, byte[] content)
{
var redPen = new Pen(Color.Red, 3);
var font = new Font(FontFamily.GenericSansSerif, 8, FontStyle.Regular);
using (var ms = new MemoryStream(content))
{
using (var img = new Bitmap(ms))
{
using (var graphics = Graphics.FromImage(img))
{
graphics.TextRenderingHint = System.Drawing.Text.TextRenderingHint.ClearTypeGridFit;
foreach (var predict in customVisionResponse.Predictions)
{
var x = predict.BoundingBox.Left * img.Width;
var y = predict.BoundingBox.Top * img.Height;
var width = predict.BoundingBox.Width * img.Width;
var height = predict.BoundingBox.Height * img.Height;
var rect = new Rectangle(
(int)(x),
(int)(y),
(int)(width),
(int)(height));
// Prediction box
graphics.DrawRectangle(redPen, rect);
// Set up string.
string measureString = $"{predict.TagName}: {predict.Probability:P1}";
// Set maximum layout size.
SizeF layoutSize = new SizeF((float)width, (float)height);
// Measure string.
SizeF stringSize = new SizeF();
stringSize = graphics.MeasureString(measureString, font, layoutSize);
// Draw rectangle representing size of string.
graphics.FillRectangle(Brushes.Red, (int)x, (int)(y - stringSize.Width * 0.2), stringSize.Width, stringSize.Height);
// Draw string to screen.
graphics.DrawString(measureString, font, Brushes.White, new PointF((float)x, (float)(y - stringSize.Width * 0.15)));
}
using (var newMS = new MemoryStream())
{
img.Save(newMS, img.RawFormat);
byte[] imageBytes = newMS.ToArray();
var base64String = Convert.ToBase64String(imageBytes);
customVisionResponse.ImageInBase64 = base64String;
}
return(customVisionResponse);
}
}
}
}
public async Task <CustomVisionResponse> PredictFromURL(string imageFileUrlPath)
{
if (string.IsNullOrWhiteSpace(imageFileUrlPath))
{
return(null);
}
try
{
var stopWatch = new Stopwatch();
HttpResponseMessage response;
var customVisionResponse = new CustomVisionResponse();
byte[] byteData = null;
using (var client = new HttpClient())
{
client.DefaultRequestHeaders.Add("Prediction-Key", predictionKey);
var url = predictionFromURL;
var request = new CustomVisionRequest {
Url = imageFileUrlPath
};
var json = JsonConvert.SerializeObject(request);
var content = new StringContent(json, Encoding.UTF8, "application/json");
stopWatch.Start();
response = await client.PostAsync(url, content).ConfigureAwait(false);
stopWatch.Stop();
}
string responseContent = await response.Content.ReadAsStringAsync().ConfigureAwait(false);
if (!response.IsSuccessStatusCode)
{
customVisionResponse.Error = $"Error {response.StatusCode}: {responseContent}";
return(customVisionResponse);
}
var tempCustomVisionResponse = JsonConvert.DeserializeObject <CustomVisionResponse>(responseContent);
customVisionResponse.Id = tempCustomVisionResponse.Id;
customVisionResponse.Created = tempCustomVisionResponse.Created;
customVisionResponse.Iteration = tempCustomVisionResponse.Iteration;
customVisionResponse.Project = tempCustomVisionResponse.Project;
customVisionResponse.Predictions = new List <Prediction>();
foreach (var prediction in tempCustomVisionResponse.Predictions.Where(x => x.Probability >= minimumThreshold))
{
var objPrediction = new Prediction();
objPrediction.TagId = prediction.TagId;
objPrediction.TagName = prediction.TagName;
objPrediction.Probability = prediction.Probability;
objPrediction.BoundingBox = new BoundingBox
{
Height = prediction.BoundingBox.Height,
Width = prediction.BoundingBox.Width,
Left = prediction.BoundingBox.Left,
Top = prediction.BoundingBox.Top
};
customVisionResponse.Predictions.Add(objPrediction);
}
byteData = await GetImageAsByteArrayAsync(imageFileUrlPath).ConfigureAwait(false);
customVisionResponse = AddCustomVisionResponseToImage(customVisionResponse, byteData);
var ts = stopWatch.Elapsed;
string elapsedTime = string.Format("{0:00}:{1:00}:{2:00}", ts.Hours, ts.Minutes, ts.Seconds);
customVisionResponse.ElapsedTime = elapsedTime;
return(customVisionResponse);
}
catch (Exception e)
{
return(null);
}
}