Improvements to Comparison, Detection, Labeler, and Debugger features

SUALAB has made another leap forward. By releasing 'SuaKIT' version 2.0, it has significantly improved upon the inconveniences of the previous version.

Version 1.0 of SuaKIT, a deep learning-based machine vision inspection software, introduced automated defect inspection to various manufacturing sectors by applying a case-learning deep learning algorithm. It has been praised for enabling inspection of various areas that were difficult to inspect with existing machine vision systems, and for significantly improving accuracy and speed.

In addition, SuaKIT 2.0 has taken it a step further by significantly improving four features: ▲ an Image Comparison mode that analyzes the difference between two images, ▲ a Detection mode that can detect and classify multiple objects within a single image, ▲ a Visual Labeler function in which a deep learning algorithm recommends defective areas of a product, and ▲ a Visual Debugger function that shows which area of an object the deep learning algorithm focused on for inspection.

Image Comparison mode facilitates optimization tailored to changes in the product or background pattern to be inspected. Naturally, it minimizes the optimization costs that were previously added.

Detection mode is a feature that can separate each type of object when multiple types are mixed within a single image, making classification difficult. It is also possible to count the number of objects.

The Visual Labeler feature is also noteworthy. Previously, defective areas on the images of every product to be inspected had to be manually specified, but in this version, a deep learning algorithm automatically recommends defective areas. This allows for the minimization of labeling costs.

Finally, there is a Visual Debugger feature. In previous versions, it was not possible to know the criteria by which the deep learning network identified defects, but now it visualizes the areas that the deep learning algorithm focused on inspecting and informs the user. This allows users to verify whether the inspection was performed according to their intent.

"SuaKIT 2.0 is the result of actively reflecting the voices of consumers and market trends," said Song Ki-young, CEO of Sua Lab. "We will continue to constantly consider ways to improve the accuracy and speed of testing and maximize user convenience."