threshold X jupyter - output folder same as input_images folder has some problem

[rajavishah]

Describe the bug

If the output and input folders are given same for threshold X jupyter then it reanalyzes the generated_output image again and produces outputs twice.

Have you searched existing issues?

• [X] I have searched and found no existing issues

Reproduction

Run the threshold_inference_jupyter_image and keep the value of --save_dir same as --folder

Screenshot

Logs/StackTrace

My Input Image Path:  /bilayers/input_images/example_input_image_output.tiff
/bilayers/threshold.py:54: UserWarning: /bilayers/input_images/example_input_image_output_output.tiff is a low contrast image
  skimage.io.imsave(output_filename, labeled_image)
Output saved to: /bilayers/input_images/example_input_image_output_output.tiff
My Input Image Path:  /bilayers/input_images/newwsr_output.tiff
/bilayers/threshold.py:54: UserWarning: /bilayers/input_images/newwsr_output_output.tiff is a low contrast image
  skimage.io.imsave(output_filename, labeled_image)
Output saved to: /bilayers/input_images/newwsr_output_output.tiff
My Input Image Path:  /bilayers/input_images/newwsr.tif
/bilayers/threshold.py:54: UserWarning: /bilayers/input_images/newwsr_output.tiff is a low contrast image
  skimage.io.imsave(output_filename, labeled_image)
Output saved to: /bilayers/input_images/newwsr_output.tiff
My Input Image Path:  /bilayers/input_images/example_input_image.tiff
/bilayers/threshold.py:54: UserWarning: /bilayers/input_images/example_input_image_output.tiff is a low contrast image
  skimage.io.imsave(output_filename, labeled_image)
Output saved to: /bilayers/input_images/example_input_image_output.tiff
Output saved to: /bilayers/input_images/example_input_image_output_output.tiff
Output saved to: /bilayers/input_images/newwsr_output_output.tiff
Output saved to: /bilayers/input_images/newwsr_output.tiff
Output saved to: /bilayers/input_images/example_input_image_output.tiff

Bilayers Version

0.1

System Info

OS version: Mac OS 14.1

[rajavishah]

It is fixed by eliminating the analysis of files with filename _output.

This behavior is also found whilst analyzing images via cellpose algorithm. So, when files named with cp_masks are passed to cellpose algorithm, It doesn't even identify it as a input_image. Following behavior was found when testing 2 files with cellpose algorithm and threshold algorithm.

---

Testing : threshold after limiting _output from processing [4 test cases] Input: file_names —> hello.tif, xyz_output.tif

• Gradio x Same_output_folder_name : outputs —> hello_output.tiff. It didn't process anything on xyz_output.tiff. Hence, returned 3 images. 2 original inputs and hello_output.tif
• Gradio x Different_output_folder_name : outputs —> hello_output.tiff. Worked as expected
• Jupyter x Same_output_folder_name : outputs —> hello_output.tiff. original 2 images were also present in the folder [Validated with cellpose’s behavior]
• Jupyter x Different_output_folder_name : outputs —> hello_output.tiff. Input images were not present in the output folder [Validated with cellpose’s behavior]

---

Testing : cellpose with file_names that includes cp_masks from processing [4 test cases] Input: file_names —> hello.tif, xyz_cp_masks.tif

• Gradio x Same_output_folder _name: outputs —> hello.tif, xyz_cp_masks.tif, hello_cp_masks.tif, hello_seg.npy. Hence, it didn't process anything on xyz_cp_masks.tiff. Returned 4 files. 2 original and 2 extra outputs of hello.tif
• Gradio x Different_output_folder_name : outputs —> hello_cp_masks.png. Was expected. Also if you choose pretrained model as cyto2, get extra outputs such as flows, png, etc.
• Jupyter x Same_output_folder_name : outputs —> xyz_cp_masks.tif was not analyzed at all. there were n different outputs from hello.tif (depending on the parameters)
• Jupyter x Different_output_folder_name : outputs —> only identifies 1 image (the message before it starts analyzing). Throws outputs for hello.tif (depending on the number of selected parameters).