Classify visually similar, but different objects while tolerating variation
The Green Classify tool is a robust classifier that distinguishes between different types of products or objects, identifies defect types, and even inspects images. Learning from a collection of labeled images, Green Classify identifies and sorts products into classes based on their common characteristics such as color, texture, materials, packaging, and defect type. This image analysis tool tolerates natural deviation within the same class and reliably distinguishes acceptable variation from different classes.
Distinguish tolerable abnormalities from true defects
Once a potential defect is spotted by the Red Analyze tool, it can be passed to the Green Classify tool to determine if it represents a true defect or an acceptable part variation. Depending on the application, not every abnormality is considered a defect. For instance, dust or discolorations might be detected, but are not considered fatal product flaws. These can be sent to a rework station rather than being discarded. By distinguishing true defects from acceptable variations, the Green Classify tool helps reduce scrap rates and control costs.
Classify defects for process control
When abnormalities are detected, the Green Classify tool can also help determine the type of defect they represent. Stains, scratches, broken elements, improper welding, and others, have different root causes. By classifying the type of defect, the Green Classify tool gives users the information needed to take corrective action and minimize continuing production of bad parts.
Choose the best tool mode depending on application needs
The Green Classify tool can be used on two different settings: focused mode or high detail mode. Focused mode is ideal for simpler applications where robust results are needed quickly. High detail mode is a powerhouse classifier for the most challenging and complex applications, which demand maximum accuracy. High detail mode also comes with a visual feedback feature to help further tune the neural network.
Users can change between modes without re-labeling images. Unlike most deep learning-based software, which require thousands of training images, the Green Classify tool trains on tens to hundreds of images in just minutes. Focus mode is good for simple applications that require a smaller set of images, a shorter training period, and run time. High detail mode is good for complex applications that require a relatively larger set of training images. Both modes can be further refined with additional images, directly on the factory floor, until it segments classes perfectly.