Mech-DLK 2.5.3 Release Notes
This section introduces the new features and improvements of Mech-DLK 2.5.3.
New Features
Added the Super-labeling Feature
You can import new images to the current module and use the Super-labeling Tool to label the images in batches. You can fine-tune the results after labeling is completed.
-
Supported modules: Instance Segmentation and Object Detection.
-
Prerequisites: The available GPU memory on the current device must be greater than 6 GB; otherwise, the model conversion may fail.
-
The Super-labeling Tool can be used only on the following three types of data:
-
Unlabeled data
-
Automatically labeled data (images with a yellow triangle at the front of the image number)
-
Manually fine-tuned data after automatic labeling (images with a yellow triangle at the front of the image number)
-
-
Three methods to use the Super-labeling Tool:
-
The Super-labeling Tool in the labeling toolbar
-
The Super-label button in the upper part of the image list
-
The Super-label option in the right-click menu
For more information, see Super-labeling.
-
Improvements
Supported the Mask Tools
Mech-DLK 2.5.3 supported the use of Mask Polygon Tool, Mask Brush Tool, Mask Lasso Tool, and Mask Eraser Tool in some algorithm modules.
-
Modules that now support mask tools: Unsupervised Segmentation, Instance Segmentation, and Object Detection.
Supported Configuring the Mask Type
Mech-DLK 2.5.3 supported configuring the mask type as “Mask globally” or “Mask single image” in some algorithm modules.
-
Modules that now support mask type configuration: Instance Segmentation and Object Detection.
Optimized the ROI Settings
Mech-DLK 2.5.3 optimized the settings for the ROI tool. After you adjust the ROI, click the OK button in the lower right corner of the ROI to save it.
Optimized the Threshold for Unsupervised Segmentation
Mech-DLK 2.5.3 optimized the default threshold position displayed in the Unsupervised Segmentation module. When the OK and NG curves do not intersect, the default threshold is set to the average of the OK and NG thresholds.
Deleted Some Parameters in Defect Segmentation
-
The "Training Parameter Settings>Training parameters>Model type>Enhanced" parameter is deleted from the Defect Segmentation module in Mech-DLK 2.5.3.
-
In Developer Mode, the "Training Parameter Settings>Data augmentation>Dilation" parameter is deleted from the Defect Segmentation module in Mech-DLK 2.5.3.