PEKAT VISION© is a software for industrial visual inspection based on deep learning
PEKAT VISION© is able to check products or input raw materials which have an unstable look or shape.
It can detect defects, find anomalies and classify such materials where current vision systems fail. It does not matter that each product/material looks different. It does not matter that it is difficult or even impossible to specify what it means when a product/material is good or defective.
PEKAT VISION© can evaluate e.g. wood, stone, rubber, leather or textile. It can evaluate agricultural products, quality of welds, surface of metal castings etc.
- The software learns by the collection of annotated images - e.g. images with positive-negative annotation. Size of the collection depends on complexity of the task.
- For collecting images to such a library the PEKAT VISION© has a feature called Auto-Trigger. With its trivial setup, it automatically captures images in the right moment without any necessary integration with the machine or assembly line.
- During the learning phase the PEKAT VISION© generalizes and understands the product/material. Afterwards it is able to independently classify, find defects and evaluate the object.
It is a new kind of visual inspection software.