Concepts in Maximo Visual Inspection Edge
Inspections are the central configuration components in Maximo® Visual Inspection Edge.
Familiarize yourself with the following concepts before you get started with Maximo Visual Inspection Edge:
- Collecting mode
- An inspection that is configured in Collecting mode collects images from an input source and sends them to a specified training data set in Maximo Visual Inspection. You can use these images to retrain a model and improve its accuracy.
- Confidence score
- A percentage value between 0 and 1 that indicates how confident a model is that a detected object is accurately identified.
A score that is close to 1 indicates that the model is confident that the object is accurately identified. A score that is close to 0 indicates that the model is not confident that the object is accurately identified. For example, a confidence score of 0.75 indicates that the model has a confidence of 75% that the object is identified correctly.
- Inferencing
- The process of running live data through a trained AI model to get a result. A trained model analyzes the images that are collected from an input source. The model returns inference results about the images, such as object names and confidence scores.
- Inspections
-
Inspections are the central configuration component in Maximo Visual Inspection Edge. Configure inspections to collect images to train models or inspect images to identify objects. You can also create inspection rules that define whether identified objects pass or fail inspections.
To configure an inspection, you must add an input source, create a station, and then select an inspection mode. For more information, see Getting started for users.
If the inspection contains rules, each inspected image includes an inspection result, which can be a pass, a fail, or inconclusive. You can view image metadata, filter images, and reset incorrect results.
The dashboard displays high-level statistics about inspections, such as the total number of passes, fails, and inconclusive results.
The integration between Maximo Visual Inspection Edge and Maximo Visual Inspection facilitates model retraining. In Maximo Visual Inspection Edge, you can use the dashboard to review the performance of inspections, verify their rule configuration and image quality, correct invalid inspection results, and prompt your model builders to retrain models in Maximo Visual Inspection.
- Inspecting mode
- An inspection that is configured in Inspecting mode uses a trained model to analyze the images that are generated from an input source. The images that are generated from the inspection results are sent to a specified inspection data set in Maximo Visual Inspection.
Because this type of inspection is associated with a trained model, the images are labeled. You can use these images to retrain a model and improve its accuracy.
You can configure rules to define whether the identified objects pass or fail an inspection. You can also send alert messages to external recipients.
For more information, see Alert messages.
- Stations
- A station is a group of inspections that logically belong together, such as a station for inspections that are located in the same place on a production line.
When you create a station, you can configure an inspection. A station can contain many inspections, but each inspection can belong to only one station.
- Input source
- An input source is the source of the images that are processed by an inspection, such as a fixed IP camera or network folder.
An inspection must have an input source so that it can receive and process data as part of an inspection. An unlimited number of inspections can use the same input source. When an input sources is added, it is available to all users.
- Models
- A set of tuned algorithms that produces a predicted output. Models are trained based on the input that is provided by a data set to classify images or video frames or find objects in images or video frames.
- Rules
- Compare inference data to configurable thresholds and define whether
the identified objects pass or fail inspections.
You can create alert messages and specify where you want to send them if the rule conditions are met. For example, you might create a message that contains instructions to stop an assembly line and specify the phone number of a line supervisor.