Traditional deep learning requires a lot of image training to get started. The computational requirements involved in training and executing the model are very high, but not all projects require such clear refinement. The Edge Learning technology introduced by Cognex allows you to learn faster using fewer images, and because of the fast learning speed, no GPU is required.

Automated visual inspection is essential to improve manufacturing speed and accuracy, so deep learning is an excellent solution. However, to effectively use deep learning technology requires a lot of image training and model execution up front, and automation engineers also need deep learning expertise. Edge learning is a viable automated solution for everyone, requiring only a few images to be trained in a short amount of time and requiring no expertise in the field to be deployed.

What is marginal learning

The so-called edge learning refers to “edge deep learning”, which is to embed efficient rule-based machine vision into a pre-trained deep learning algorithm to create an integrated toolset optimized for factory automation. Edge Learning can be deployed on any production line in minutes using a solution based on a single smart camera. Edge Learning differs from other deep learning products in that it focuses on ensuring ease of use at all stages of application deployment. For example, edge learning requires fewer images for proof of concept, takes less time to set up and capture images, and requires no special programming.

3 advantages of edge learning

1. No experience required

The technology requires no expertise in machine vision or deep learning. Instead, line engineers can train edge learning techniques based on their existing knowledge of the tasks they need to solve.

2. Easy to deploy

By using a solution based on a single smart camera, users can deploy edge learning on any production line in minutes. The solution integrates high-quality visual hardware, machine vision tools used to pre-process each image to reduce computational effort, deep learning networks pre-trained to solve factory automation problems, and a simple user interface designed for industrial applications.

3, easy to use

Edge learning is not a universal solution, but is specifically tailored for industrial automation applications. Edge Learning differs from other deep learning products in that it focuses on ensuring ease of use at all stages of application deployment. Compared to more traditional deep learning solutions, the technique is simple to set up, requires less time and graphics to train, and requires no programming experience.