Developing an edge AI application is a big task. In this chapter, we’ll get familiar with the iterative development model that helps deliver successful edge AI deployments in real-world projects.
An Iterative Workflow for Edge AI Development
The process of developing a successful application is fundamentally simple: start small, make incremental changes, measure your progress, and quit when you meet your goals. The complexity comes when you introduce the vast number of moving parts that make up the technology of edge AI. This section of the book aims to provide a concrete process you can walk through to maximize your chances of success.
As we heard back in “The Edge AI Workflow”, the core idea behind this workflow is the power of feedback loops. Our goal is to create feedback loops between the various stages of the process, leading to an ever-improving understanding of the problem, our solution, and the best ways to fit them together (as shown in Figure 9-1).
While it’s an iterative process, some parts are more iterative than others. The steps we tackle earliest—exploration, goal setting, and bootstrapping—are the parts where we’re figuring what we want to do and how we may be able to go about doing it. They feature first in up-front planning and then during periodic reappraisal, as new information comes in: perhaps after an initial deployment, or when a significant amount of new data has come to light.
Figure 9-1. Feedback loops are ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month, and much more.
