Real-Time Image Processing with Oakestra Framework (Complete Theory)

Prachi Rawat

This document explains how real-time image processing can be performed using the Oakestra orchestration framework, specifically using Raspberry Pi as an edge (worker) node. It includes complete theory, workflow, internal scheduler behavior, and expected monitoring outcomes without requiring actual implementation.Deploy an image processing service (e.g., object detection) at the edge using Oakestra, where:

  1. Job Request Creation: The local software creates an SLA describing resource needs.
  2. Root Scheduler (Cloud_Scheduler): Receives SLA, selects best-fit cluster using Celery + Redis + Resource Abstractor.
  3. Cluster Scheduler (Cluster_Scheduler): Filters eligible nodes, scores them, and chooses the best one.
  4. Service Deployment: Node Engine deploys the container to Raspberry Pi.
  5. Inference and Output: Raspberry Pi processes the image and sends results to the system.