User Defined Pipeline

Use a User Defined Pipeline (Pipeline) skill to create custom, engineered workflows by using out-of-the-box blocks to identify or achieve a specific image related to anomaly detection. You can configure and place the blocks in a specific order. You can then train the Pipeline to detect an anomaly. A Pipeline skill allows you to implement automated steps like the following example workflow:

1. Capture an image.

2. Set the focus on a key part of the image.

3. Zoom, rotate, or align the image based on specific elements.

4. Run the output through a range of algorithms:

   1. Perform a measurement (such as length, gap width, or amount of deviation from an expected shape).

   2. Detect an anomaly.

   3. Record the presence or absence of a known item (for example, a hardhat or vest).

   4. Send results to the appropriate destinations.

For example, consider a visual parts inspection scenario where an object is inspected to determine any defects, such as scratches, dirt, foreign material, or any other physical defect. You can automate this inspection by using a Pipeline skill. In the Pipeline skill, you can use the image captured by a camera (at a particular point and angle) and apply machine learning or deep learning algorithms to determine if there is a defect.