Running Machine Vision in the Observation Scan module

When the Images have been reviewed and prepared it is time to run Machine Vision. Machine Vision makes use of the AIR algorithm to identify integrity issues for the provided Images.

Selecting the Tags

First you need to decide on which set of Images you would like to run Machine Vision. Machine Vision can be run on separate Batches, or on one (or more) of the Observation Scan subfolders, or on the entire Observation Scan project.

Splitting into Batches

Any "last level" folder (that has not already been split into Batches) can be split into Batches.

To split a Tag into batches:

Select one Tag from the Observation Scan List. This has to be a "last level" folder that has not yet been split up into Batches.
Click the Split button.
Select a last level Tag and click the Split button.
- Specify the Number of batches or Number of Images per batch.
- Click Split.

Machine Vision Settings

To edit the default Observation Scan settings:

Ensure that no Tags are selected.
Click on the Observation Scan Settings button.
Click Edit.

Select the desired defaults for Observation Scan.
Click the Save button.

Editing Default Observation Scan Settings.

To edit the Observation Scan settings for one or more Tags:

Select the desired Tag.
Click on the Observation Scan Settings button.
Observation Scan 3D Settings.
Click Edit (or Add if no settings are set for the Tag).

Now you can edit the desired settings, Issue Classes and Filter Image and Segmentation Factor.
When done, click Save.

Machine Vision Stages

Machine Vision runs in one to three stages, depending on the Filter Images setting and the Segmentation factor value in the Observation Scan settings.

The three stages are:

FILTERING – In this stages similar Images are filtered. This stage is performed only if the Filter Images setting is turned on in the Observation Scan Settings.
FILTERING PATCHES – This stage filters out similar Image patches. It is performed only if the Segmentation factor in the Observation settings is greater than 1 and the Filter Images setting is on, i.e., checked, in the Observation Scan Settings.
MACHINE VISION – Issues detection.

The Filter Images setting will determine if the results of previous Machine Vision runs will be reused. When Filter Images is off, i.e., unchecked, Machine Vision will skip previously analyzed Images. If Filter Images is on, i.e., checked, all previous Machine Vision results will be deleted and Machine Vision will be run again.

The Segmentation factor setting should be used when high resolution Images are provided. Increasing the Segmentation factor will divide large Images into smaller ones. For example, with the Segmentation factor set to 4, Images will be divided into 16 patches (4 x 4), with each patch being treated as a separate Image. Machine Vision returns better results when high resolution Images are split into smaller Images.

Selecting which Issue Classes to report

Only the Issue Classes selected in the settings will be reported. If the settings has not been defined for a Tag, it will inherit the settings from the level above.

To select the Issue Classes:

Select the desired Tag.
Open the Observation Scan setting and click Edit / Add.
Select the desired Issue Classes. Take note that some Issue Classes are grouped together. You can expand these to see the individual Issue Classes.
Select the Issue Classes.
Click Save.

The settings are applied only to the selected Tags.

Running Machine Vision

To run Machine Vision:

Select at least one Tag from the Observation Scan List.
Click the Run Machine Vision button.

Selecting and Running Machine Vision Jobs.

After clicking the Run Machine Vision button, a Machine Vision Approval popup will appear.

Enter the justification or notes associated with your authorization request.
Click Send.

IMS will send an approval request by email to the authorized person with the details of the Machine Vision Run.

When the authorized person clicks the link included in the email, the Machine Vision Approval window will appear.

As an authorized person, you can either Approve or Disapprove the Machine Vision request. You can also send a message to requester.

Click Save to approve or disapprove the request. If the request has been Disapproved, an email will be sent to the requester with any notes added to the approval request.

Machine Vision Job Status

You can keep track of the progress of the Machine Vision jobs in the MV Progress column.

Table Description automatically generated — Machine Vision Progress.

Here you will see in which Stage it currently is in:

PENDING APPROVAL: The Machine Vision Job request has been submitted and is waiting approval from the Authorized Person.
DISAPPROVED: The Machine Vision Job request has been rejected by the Authorized Person.
FILTERING: The Machine Vision Job request has been approved and is being processed.

If needed you can stop the Machine Vision run. To do so click the Stop Machine Vision button in the toolbar.