Edge Learning Overview

Edge learning brain icon against teal background with connected nodes

Edge Learning Basics

Edge learning is a subset of artificial intelligence (AI) in which processing takes place on-device, or “at the edge” of where the data originates, using a pre-trained set of algorithms. The technology is simple to setup, requiring less time and fewer images for training compared to other AI-based solutions, like deep learning.

Edge learning is the answer for both engineers looking for an easy way to integrate automation into their lines and for expert automation engineers who regularly use rule-based machine vision tools but lack specific AI or deep learning expertise. This makes the technology a viable solution for all—from machine vision beginners to experts – to solve a range of applications across the factory and across industries.

Example use case: classification

Edge learning is powerful enough to analyze multiple regions of interest (ROIs) in its field of view and classify each of those regions into multiple categories. This enables users to perform sophisticated assembly verification.

For example, edge learning can verify and sort four sections of a frozen meal tray on a high-speed line. In each tray, the bottom center section contains the protein, the top left the vegetable, the top middle the dessert or side dish, and the top right the starch. Each of the sections can contain multiple SKUs, like chicken, turkey, or meatloaf in the protein section, and rice, potatoes, or pasta in the starch section.

With a simple click and drag, each region can be defined and locked to invariant features on the meal tray. After that, the edge learning tool is trained to classify each section of the tray with only a handful of images, often as few as two for each possible class. Within minutes of training, edge learning will accurately classify the different sections at high speeds. If more variation is introduced, for instance either a new class or a new option within the same class, the edge learning tool can be updated with a few images of the new category.

What works for frozen meal trays also works for classifying parts and products across a range of industries as seen in the edge learning application examples.

Edge learning inspects frozen food tray by detecting and classifying different sections

Advantages of Using Edge Learning over Deep Learning

Edge learning allows you to combine efficient rule-based machine vision within a set of pre-trained AI algorithms to create an integrated toolset optimized for factory automation. This technology requires neither specialized knowledge of machine vision nor AI. Instead, line engineers can train edge learning using their existing knowledge of required tasks. This makes the technology a viable automation solution for all—from machine vision beginners to experts. Read on to learn more about the benefits of using edge learning in your operations.

	Deep Learning		Edge Learning	Advantage
Hundreds to thousands of images required for training		Five to ten images required for training		Fewer images required for training
Hours to days required for processing		Seconds to minutes required for learning		Faster learning
Significant understanding of deep learning systems and programming needed		No prior experience needed		Higher ease of use