kahst / BirdNET-Analyzer

BirdNET analyzer for scientific audio data processing.
Other
870 stars 158 forks source link

Review-Fuction: Regression dialogue, Precision disapears after last reviewed segment of species #493

Open Richterskala101 opened 1 month ago

Richterskala101 commented 1 month ago

Describe the bug A clear and concise description of what the bug is. When reviewing segments with the review function of the Birdnet-Analyzer, after being done with the last segment of one species/class, the small box in the regression-window showing the precision values for different thresholds disappears. That happens, once you are done with all the segments available for a species/class. Then the little box on the right, down at the "Regression" part, disappears. I am unsure if that behaviour is on purpose. Additionally, it would be very convenient, to save these results in some form of table. As I understand it now, one must at least screenshot the result, to save it for later.

To Reproduce Steps to reproduce the behaviour: For the GUI:

  1. Go to Review
  2. Click on select input directory
  3. choose a directory with segments extracted in advance
  4. review segments for a given species
  5. see the "Regression" pictured below, with the precision values for different thresholds
  6. review until you reach the last segment of the given species
  7. In the "Regression" figure, see that the precision values for different thresholds are gone now. For the CLI provide us with the arguments used.

Expected behavior In "Regression" figure, t the precision-values for different thresholds should remain, after review of last segment.

Desktop:

max-mauermann commented 3 weeks ago

This might happen because we hide the threshold values if the are > 1. So if the last review-segment pushes all the thresholds above 1 the box will disappear. Otherwise i could not reproduce this issue.

Saving the result as a table is a good idea. (Other then screenshotting the result you can also reopen the folder with the BirdNET GUI to access the results, but i see that this is not really convenient)