Deployed model location

Locally deployed models

Running an inspection with a locally deployed model is more efficient than a remotely deployed model because:

• There is less network traffic.

  The inferencing of images and the generation of inspection rules and alerts are conducted locally on Maximo Visual Inspection Edge. Then, the final results are sent to the Maximo Visual Inspection training server. Less network traffic between Maximo Visual Inspection and Maximo Visual Inspection Edge increases efficiency.

• During a loss of network connectivity, an inspection continues to run.

  When the network connection is restored, the collected inspection results are transferred to Maximo Visual Inspection with a scheduled data push that is managed through the data transfer policy. For more information, see Data transfer policy.

Remotely deployed models

Running an inspection with a remotely deployed model is less efficient than a locally deployed model because:

• There is more network traffic.

  Inspection images are sent to Maximo Visual Inspection for inferencing. The inference results are sent back to Maximo Visual Inspection Edge and the associated inspection rules and alerts are generated. Then, the final results are sent back to Maximo Visual Inspection. More network traffic between Maximo Visual Inspection and Maximo Visual Inspection Edge decreases efficiency.

• During a loss of network connectivity, an inspection does not continue to run until the network connection is restored.

However, using a remotely deployed model is beneficial if you do not have sufficient GPUs on your instance of Maximo Visual Inspection Edge.