Cleaning glass bottles is a significant challenge in the beverage industry
Incoming bottles are fed onto a conveyor belt and a new-glass and old-glass program are used to process the bottles. New glass requires a lower drive factor than old glass, and if the drive factor is too high, it makes the drive pressure too strong, which leads to the bottles bouncing on the feed system and the machine to stop. Selecting the appropriate program is a time-consuming manual task and has thus far required a lot of human attention.
The inspection solution consists of the InspectorP62x 2D vision camera and the Intelligent Inspection SensorApp. The stream of bottles is automatically classified into new and old glass. Example images are used to train the neural networks in the cloud. Drawing on this reference, the intelligent algorithm then evaluates images directly in the camera. This makes switching between programs for new and old glass easy.

The solution is practical and affordable
By combining traditional machine vision with deep learning, users can automate demanding inspection tasks at a low cost. Plus, the customer has total ownership, since the deep learning network runs directly on the camera without any additional hardware. And on the shop floor, increased machine availability is achieved with fewer downtimes, and water consumption is optimized because restarts are avoided.










