still high memory usage and consuption

[areytechai]

Title: High Memory Usage in PredictorPlottingExtensions Class

Description: I’m experiencing significant memory usage when using the PredictorPlottingExtensions class in the YOLOv8 project. During image processing, memory consumption increases rapidly, leading to performance issues, especially with large images.

Steps to Reproduce:

Use methods like PoseAndSave, DetectAndSave, DetectObbAndSave, SegmentAndSave, or ClassifyAndSave for image processing and saving. Observe memory usage as the application runs. Expected Behavior: Memory consumption should remain stable and efficient during image processing and saving.

Actual Behavior: Memory usage increases significantly and doesn't seem to be released, potentially indicating a memory leak or inefficient memory management.

[areytechai]

[]single i couldnt stop Single[] 99 457184344 457184344

[dme-compunet]

It was fixed in version 5.0.2, which version are you using?

[areytechai]

It was fixed in version 5.0.2, which version are you using?

The latest version, 5.04,

after a while

private static string CreateImageOutputPath(string path, YoloTask task) { var baseDirectory = Path.GetDirectoryName(path) ?? Environment.CurrentDirectory;

  var plotDirectory = Path.Combine(baseDirectory, task.ToString().ToLower());

  if (Directory.Exists(plotDirectory) == false)
  {
      Directory.CreateDirectory(plotDirectory);
  }

  var extn = Path.GetExtension(path);
  var name = Path.GetFileNameWithoutExtension(path);

  var index = 0;

  while (true)
  {
      var filename = index == 0 ? $"{name}{extn}" : $"{name}_{index}{extn}";
      var fullpath = Path.Combine(plotDirectory, filename);
      return fullpath;

      //if (File.Exists(fullpath))
      //{
      //    index++;
      //}
      //else
      //{
      //    return fullpath;
      //}
  }

}

i try to replace alltime with same file isit cause of it?

single[] is extremly high

[areytechai]

the single[] class unstopable

[dme-compunet]

According to the memory graph it seems that there is no leak, but the memory is quite high, and this is possible if you load a large model or several models at the same time, How much does the model you use weigh?

[areytechai]

just one image progressing, in 3-4 seconds only segmentation, i just make timer, after 60 minutes restart apps and free the memmory,

any alternative way to clear single[] ?