Tracing in AI Session Processor

Fully Manual

This example shows a fully manual approach to tracing features directly in the AI Session Processor.

Downloads

You may download the sample reference images to use as learning tools and follow this example.

⇩ Reference Images

Load References

Begin by launching AI Session Processor from the Launch Bar.

Drag and drop your reference images.

Add Layers

Click + in the Layers Panel to add a blank layer. Set the name and click “OK”. You may add additional Layers at any time using the + button. Click the Layer name in the Layers Table to set the active Layer for tracing.

Trace Features

Use manual editing tools to trace features of interest.

*Tip: The quality of model training will depend on tracing quality, so take care to trace features as accurately as possible. Tracing all features of interest is recommended, but not required. Pixels which are not assigned to any Layer will be ignored during training.

Click Save and advance to the next image to continue tracing.

After all features have been traced in each image, you may use the dropdown arrow next to the + followed by Add Inverse of All Layers to add a background Layer if needed. Name the Layer and click “OK”. You’ll usually choose “No” when asked if you want to erode after inversion. This will add all undefined pixels as another Layer.

*Tip: Training requires at least 2 Layers. Only use the above inversion method if all features in Layer 1 are traced. If they are not, click + to add the second Layer, and trace a few areas of the background.

Semi-automated

This example shows a semi-automated approach to tracing features in the AI Session Processor. This approach is useful if you have a Recipe which does a reasonable job of detecting features of interest, but leaves room for improvement

⇩ Download example Recipe and image

Batch Apply

Launch the Batch Processor.

Load Recipe and drag and drop images to batch process.

Click Process.

Clean Up

When batch is finished, launch AI Session Processor.

Drag and drop the Session file which was saved to the output directory of the batch process.

Use manual edit tools to correct any detection errors in all Layers of the first image.

Click Save and advance to the next image to continue correcting. Repeat until all images are corrected.