The inference section includes the pictures which were saved by the system on its own while the model was being used. The images are tagged with objects which were identified by the model. Users can also correct them here.
Inference can be used to see if the model is identifying images correctly on its own or not.
To be able to access the Inference page, you need to have the following enabled on the FCC side:
- Device Flow (keycloak) enabled
- use the following property for the Frontline Workplace
App: AI: Inference: SaveImages: true
Note: Please reach out to your Customer Success Manager to assist you on how to configure Frontline Workplace.
Multiple images can be selected by pressing Ctrl while clicking on them. The images can be untagged or edited after being selected.
Note: Similar to Dataset section, the images present in Inference can also be sorted and filtered. They can also be filtered by selecting the minimum confidence level on the left.