Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
Create custom insights to detect objects or situations tailored to your operations without coding. Use natural language descriptions, images, and examples to build detections for anything from safety equipment to workflow conditions. Then apply them instantly to live camera streams.
Prerequisites
Before you begin:
- Review the Custom Insights overview to understand the best practices and limitations.
- Review limitations for all live AI insight detections.
Create a custom AI insight
You can create a custom AI real-time insight in the VI portal to detect either objects or situations. The steps are mostly the same for both types, with a few differences noted below.
To create a new custom AI real-time insight:
Go to your VI extension → Manage AI insight.
Move to the AI insights tab.
Select Create custom insight.
Choose the type of AI insight you want to create: Object or Situation.
Enter a name for the insight in the AI insight name field. The name isn't part of the training data.
For object detection:
- Detected object: Describe the object to detect.
- Object example images: Upload up to 10 images of the object.
- Select the Fine-tune (optional) tab to add negative examples.
- Add words to name objects you don't want to detect. Separate the words by commas.
- Upload up to 10 images that represent objects you do not want to detect.
For situation detection:
- Detected situation: Describe the situation you want to detect, such as
A long queue in a store.
- Select the Fine-tune (optional) tab to add negative examples. Describe situations you don’t want to detect. Separate examples with commas.
- Detected situation: Describe the situation you want to detect, such as
Review and select Create insight.
Related content
After you create a custom insight: