Traditional rule-based image processing is very useful for a range of inspection tasks involving uniform parts in a controlled environment. However, when it comes to, for example, solving hard-to-program image processing applications with rule-based algorithms, handling cluttered backgrounds and variations in the appearance of inspection parts, and training existing applications with new image data, Deep Learning provides a groundbreaking alternative. IMAGO’s box PCs and intelligent cameras become powerful inference computers and cameras with a wide range of applications when a neural network is applied. I.e., they can apply their “knowledge” acquired through Deep Learning to new data. Neural networks are executed directly on the devices in a hardware-accelerated manner. This enables inference times of a few milliseconds.
However, IMAGO’s models are not only intelligent, but also consistently suitable for industrial use: With features such as C-mount, robust housing, GigE network connection with RJ45 GigE connector, RS-232 interface and REST web interface, customers receive fully-fledged industrial cameras and computers that prove themselves even in demanding environmental conditions.