Manual

Table Of Contents

DobotVisionStudio User Guide

150

For defects, the pixel value is high, and most of the interference can be removed through subsequent

processing or secondary training. The image on the right side of the above figure is the defect

probability diagram of image segmentation.

• Multi-level Classification Mode

For defects, the pixel value is high, and most of the interference can be removed through subsequent

processing or secondary training. Through subsequent processing, the image on the right side of the

above figure is the defect probability diagram of image segmentation.

• Multi-level Classification Mode

The right side of the figure above is the category diagram of image segmentation, and its operation

parameter settings are shown in the figure below.

Running Parameters of Defect Detection

Model File Path

Select the model file generated by the previous image segmentation training.

Saving Model in Solution

After enabling, save the model data to the solution file or process file. When

loading a solution across machines, you do not need to enter the model file path.

Min. Score

It is classified as a certain type of probability.

Display Probability Diagram

You can select defect probability diagram displayed on the display interface.

Output Type

You can select the defect probability diagram that can be subscribed by subsequent

modules.

NOTE

 The DL image segmentation module supports the multi classification function of

defects. When marking defects with the deep learning training tool, if it is necessary

to classify defects, mark the defect type at the same time, and only the large/small

model of single image segmentation and the self-learning template mode in image

comparison support the multi classification function.

DL Classification

Deep learning classification is an image processing method for distinguishing the targets of different

types according to different features reflected in image information. It uses the PC to analyze images,