Can defects detected by vision systems, or how we solve your production problems

Quality systems in the food industry based on machine vision are a proven and extremely effective solution. Up to 300 cans per minute can be inspected by our advanced vision systems – each can is checked with full precision. Despite production challenges (line speed), Scanway’s machine vision systems guarantee accuracy, detect defects effectively, ensuring 100% inspection of each can without slowing down production.

	photo from a system with a double seam defect
	photo of a system with a dent defect
	a snapshot of the system with key presence and orientation control
	system photo with expiration date control

What disadvantages of cans are eliminated by the vision system:

1. Seam defects: teeth, drooping, dents, creases not connected to the body, excess material protruding beyond the edge, seam breakage and cuts, cut collar, seam jamming.
2. Disturbed geometry: dents and the like.
3. Defective printing: incomplete printing, printing inconsistent with the batch, inconsistency of the lid with the bottom of the can.
4. Label errors: missing, unstuck or misplaced.
5. Key defects: no key, shifts.

Benefits of implementing CUMULUS vision systems for quality control of metal cans in the food industry

Advantages of automated quality control based on machine vision

Above all, vision-based quality control is many times faster and more accurate than manual inspection. With its help, each of the 300 cans moving on the production line’s belt can be checked in one minute. Thanks to integration with the rejector, any defective product is immediately removed from the line. By detecting product anomalies, the CUMULUS system also makes it possible to predict potential machine failure. All data on detected defects are meticulously collected and archived, which makes it possible to conduct detailed analysis and create precise statistics.

How did we create a vision system for can inspection?

Based on a detailed analysis of the process and our experience together with customers, we have created a dedicated solution tailored to the specifics of this product. Our implementations do not disrupt already existing production processes, but complement them. This is best exemplified by an example – below is the characteristics of one of our implemented jobs:

1. Five infrared field cameras for precise canned imaging.
2. Stable illumination to allow uninterrupted operation of cameras and algorithms.
3. AI algorithms that analyze camera images in real time, classifying cans as OK/NOK.
4. A water droplet blow-off system that eliminated false NOKi associated with water lingering on the flange of the can.
5. Non-destructive mechanical rejector with buffer for NOK cans.
6. A system that allows quality data to be recorded and quickly recalibrated for other references using recipe recording.

Implementation of the system on the production line

With our image processing and analysis technologies, we significantly reduce the time to implement solutions – from over 9 months to 3-6 months. Our vision systems, based on proprietary HYDRA technology, combine various imaging techniques and proprietary algorithmic solutions. We use AI/ML for multi-threaded evaluation of the quality of manufactured products and prediction of events on production lines. In this way, we address the most important problems of modern plants – the need for continuous monitoring, optimization and automation of processes.

DOWNLOAD PDF:

– Implementation. CUMULUS group vision system for can inspection.

– CUMULUS. Vision-based quality control of a single object.