HomeDocumentation
Log In
Documentation

Inspecting with Multiple Regions of Interest

The Region of Interest (ROI) tool in Brain Builder and Inspector can focus on a specific area of each image being inspected. This can be useful when there is information in the image that is not relevant to the inspection but that confuse the Brain, such as multiple products in view or other background visibility. The ROI can also be used to focus the Brain on a specific area of a product for a more precise inspection of only that component. This can be useful when analyzing detailed areas like weld points or complex equipment such as circuit boards.

It is also possible to apply multiple regions of interest when inspecting a single product by following the instructions on this page.

Brain Training

Each Brain can only be trained with one Region of Interest. To run inspections with multiple ROIs, it is necessary to train a Brain for each region.

Keep the following details in mind when training these Brains:

  • Because the Brains will all be inspecting the same images, the images should be consistent for each Brain. The Brains do not have to be trained with the exact same images, but the images should not vary in terms of lighting, camera, or other variables.
  • Make sure that the images are labeled appropriately for each region of interest. An image that might be "anomaly" for one region could be "normal" for another.
  • Train, evaluate, optimize, and test each Brain independently to ensure optimal performance

After each Brain has been trained, Deploy the Brain within Brain Builder so that it is ready for inspections.

Setting Up Inspections

As with other inspections, the normal steps should be taken in Inspector:

  1. Import the Brains that have been trained. Configure the output for each Brain according to the details of the PLC and automation system.
  2. Add all of the Brains to the input on the Inspect page. Use the + button multiple times to associate multiple Brains with a single input.
  3. Use the Test screen to confirm each behavior is being applied as expected with the ROI.
  4. When ready, Start Inspecting. Inspector will analyze images from the input using all three Brains, and outputs will be sent for each result. The Inspect page will display a count of the number of inspections and the number of instances of each class output for each behavior.

Configuration Flexibility

Inspector supports the use of mixed Brain Types when running multiple Brains on the same image. It is also possible to mix Brains that use ROIs with those that don't. Each Brain is effectively running an independent inspection of the image and reporting the results, and that allows for wide flexibility in how the tools are deployed and used during inspections.