Visual Trigger

Visual Trigger is a mode that uses AI to run an inspection.

This mode can detect when an object comes into the camera frame, identify that object, and run an inspection. The inspection’s trigger string matches the label name.

Visual Trigger does not require an external trigger that uses MQTT, although that type of trigger can work in parallel with Visual Trigger.

When Visual Trigger finds a label in the frame that satisfies the frame timing logic, an internal trigger message is sent within the app, not by using MQTT. Through this trigger message, an attempt is made to identify an inspection that has a trigger string with a matching label. Usually a separate inspection is assessed, but it is possible to use the same inspection. However, if you use the same inspection, make sure that the trigger string on the inspection matches the label that is being searched in the frame.

Visual Trigger has the following requirements:

Although object detection and classification models are supported, a classification model must be well-trained because its confidence levels can fluctuate.
The inspection must have at least one label with frame timing configuration. You can optionally assign an obstruction label to prevent obstructions, such as a person who walks into the frame.
The label name or names of the model in the inspection must be trained to match the trigger string that is defined in the inspection.

Note: When you use Visual Trigger, the maximum image size that is uploaded is 1080 pixels.

Visual Trigger can run in parallel with Barcode and OCR Trigger.