Covision Quality automates the visual quality control process on metal parts through computer vision and deep learning technology.
Covision Quality was born out of Covision Lab, a computer vision focused research and development center with the goal of shaping industries through its applications. It was founded in 2019 by seven multinational technology and industrial companies.
In order to make visual end-of-line quality control and defect detection more reliable and scalable Covision is applying a computer vision
technique called ‘unsupervised learning’ which enables metal manufacturing and processing companies to get the software to work in a few days, without having to individually program and continuously fine-tune a vision inspection system for each metal piece.
Three aspects of the software are key to making this happen:
Unsupervised Learning: At scale, the software learns itself to classify good and bad parts, which means the solution is fast and does disrupting the production in the visual inspection process.
Transfer Learning: The software can transfer knowledge on defect detection of existing and proven deployments from one metal object to another. This means time to scale to other production lines is significantly reduced.
Continuous Learning: Algorithms get continuously improved via a continual learning approach. This means that over time and with increasing input of data from many production lines our software gets continuously improved and more reliable.
Today, end-of-line quality control is a major pain point with visual quality control relying heavily on humans to visually inspect objects with human error rates between 5 and 30%. Current vision systems are difficult to program and do not scale with programming completed manually by specialist which can take weeks per part which does not scale to multiple different parts.
Covision Quality technology can be used to automate the end-of-line quality control process in various downstream industries that leverage non-ferrous metals such as aluminium, copper, and zinc as well as ferrous materials such as steel.
For more information: www.covisionlab.com/covision-quality