Skip to content

🔍 Inspecting Masks

Data quality is of utmost importance when labeling the images.
In version 0.2.1 of DL_Track_US, we included an option to inspect labeled images and corresponding masks.


1. Accessing the Mask Inspection Tool

  • Open the UI.
  • In the main window, click on Advanced Methods.
  • In the Select Method dropdown, choose "Inspect Masks".
  • The Mask Inspection Window will open.

adv_meth select_meth inspect


2. Selecting Relevant Directories

You need to specify three directories:

  • 📁 output_images — contains the original labeled images
  • 📁 fascicle_masks — contains fascicle masks
  • 📁 aponeurosis_masks — contains aponeurosis masks

These folders should have been created during the Image Labeling process.

Fascicle and aponeurosis masks must be inspected separately.

Specify:

  • Click Image Dir to select the output_images folder.
  • Click Mask Dir to select the respective masks folder (either fascicle or aponeurosis masks).
  • Use the Start Index to choose the starting image number.

mask_button


3. Starting the Inspection

  • Click on Inspect Masks to start inspecting.

One of two things will happen:


Case 1: Everything Matches

  • Number of images and masks is equal.
  • Naming conventions are correct.
  • You will see a messagebox confirming everything is OK.

no_different

Click OK to continue to the Mask Inspection GUI.


Case 2: Mismatch Detected

  • Number of images and masks is not equal and/or
  • Naming is incorrect.

In this case:

  • A table appears showing:
  • Incorrect image names
  • Which directory they occur in
  • If the number of files differs

💡 Tip:
Adjust the files according to the table, then restart the inspection.

differen_images


4. Using the Mask Inspection UI

Once inspection starts:

inpsection

In this UI:

  • The original images are shown.
  • The masks are overlaid in semi-transparent green.
  • Follow the instructions displayed inside the GUI carefully.

🛑 Delete Warning:
Clicking the Delete button will permanently delete the selected image-mask pair!

We recommend making backups of the folders before starting inspection,
especially if you might want to correct masks later.


✅ You can now inspect, validate, and clean your datasets.