Introduction

MIPAR REST API is a powerful tool that packages the MIPAR engine into a Docker Image

Scope
The MIPAR REST API was developed to meet the needs of customers who are already using MIPAR to do powerful image analysis, but would like to setup “headless” workstations that execute algorithms with minimal supervision. The REST API is deployed using Docker as it is extremely versatile and widely adopted.

Capabilities
At this time the MIPAR REST API is only able to execute pre-built MIPAR recipes, this means that you must have an already developed MIPAR recipe to utilize the MIPAR REST API. The Docker image accepts images, recipes and returns the resulting binary masks and global measurements.

Recipe Step Compatibility The following recipe step types are not supported:

• Spotlight steps – Recipes containing Spotlight steps cannot be executed via the REST API. For headless execution, use a trained MIPAR deep learning model in place of Spotlight for the detection step.
• Imported PyTorch model steps – Deep learning steps that apply an externally imported PyTorch model (.pt file) are not supported. Only MIPAR-native .dlm models can be applied through the REST API.
• Interruptible steps – Steps configured as interruptible (requiring user input) are skipped during REST API execution.