Deep learning machine vision for glass bottles
Deep Capture, an industrial vision software using deep learning, enables the detection of bottle positions (X, Y, Theta) in various types of containers dynamically on a conveyor.
The system is integrated into a robotic depacker, allowing the bottles to be picked up for loading onto a filling line.
The challenge in detection lies in the wide variety of bottle types to be processed (around fifty) and the different containers (cardboard boxes and thermoformed trays).
Deep Capture simplifies setup and format creation. A single configuration of lighting and cameras enables detection across all bottle types. An algorithm organizes and sends the pick-up positions to the robot.
