Machine Learning: Intelligent Empty Bottle Inspectors are changing the market
The beverage and liquid-food industry is putting today’s most innovative tools to work, including machine learning. It’s a good example of the possibilities that digital transformation offers the industry. Using “deep learning software”, machinery manufacturers can integrate self-learning programs at the forefront of the filling plant to implement smarter approaches to empty bottle inspection. These innovations create more efficient and sustainable processes like:
- Machine Learning and the subfield Deep Learning offer the beverage industry technological innovations through software solutions based on artificial neural networks.
- Inspection machines “trained” using deep learning inspect empty bottles: they are thoroughly examined for defects before being introduced into bottling processes.
- Teachable machines also rely on their “training partners”: people classify information, which the system then processes further within the neural network.
Empty bottle inspectors generally use a specific technique. In order to professionally check empty bottles, inspection machines are needed to perform quality inspections on glass or PET bottles. These inspections not only identify damage like cracks, but also residual liquids or bubbles in the layers of glass. Illuminating all sides of the bottles, using high-resolution cameras, allows for high-level precision in these processes.
How does Deep Learning work in Empty Bottle Inspection?
Deep Learning technology offers an advantage in empty bottle inspection: this specific type of machine learning is based on an artificial neural network, which stores acquired data. The software is continually trained to achieve high-level accuracy when inspecting and detecting flaws. Problems detected during the process are used to “train” the software and its detection models, so if problems recur, bottles can be correctly categorized as “good” or “bad”.
Deep Learning in Detail: Teachable machines analyze individual data sets, enriching their network with more and more information. The deep-learning software begins without prior knowledge about the bottle inspection process and learns step by step, using the data provided. And it does all of this with extreme speed. High-level precision quickly reveals noteable patterns in the data, and the rate of defects is continuously minimized. Self-learning programs identify countless objects, shapes, and pixels. This means they recognize the most delicate objects during the inspection, such as drops of water.
The Advantages: Deep Learning processes are suitable for glass and PET containers, getting top marks for their reduced rate of error. Additionally, once the one-time installation takes place, the machine in the bottling plant does not need any further adjustments. After initial operation, empty bottle inspection is carried out exclusively via the self-learning system.
The Technological Experts:
- Krones AG: Linatronic AI
Their 40 years of inspection technology experience has produced the “Linatronic AI” system. Using Machine Learning, this system significantly reduces the time required to establish initial machine operation, improves bottle bottom inspection and reduces false rejection in bottling plants by up to 50%.
- HEUFT Systemtechnik GmbH: eXaminer II XOS
The HEUFT company uses x-ray images in its “eXaminer” software. Once trained, the system combines radiometric and optical recognition technologies during empty bottle inspection to improve image quality and performance in image evaluation.
The developers and technologies from Krones and HEUFT, along with many other specialists in processing, filling and packaging technology specialists, will gather from September 12 to 16 in Munich, at drinktec 2022. Check out our directory of notable and exciting exhibitors.
How future-proof is Deep Learning?
Highly Efficient: Empty bottle inspectors combined with deep learning technology conserve human resources because there is minimal need to define strict parameters. Teachable machines trained using Deep Learning are faster – and therefore contribute value to filling plants because of increased efficiency.
Highly Sustainable: Since many companies in the beverage and liquid food industry have a goal to operate in an environmentally friendly manner as possible, reducing the use of raw materials is a fundamental focus. Deep Learning software helps conserve raw materials by only rejecting bottles when they are actually defective.
What challenges do smart Empty Bottle Inspection Machines face?
Teachable machines need support too. Even with self-learning programs, human technical know-how is still necessary.
As always, for the first, one-time technical set-up of the inspector, a specialist – such as an experienced electrician – is always required.
For collecting and capturing data sets extensively, a human perspective is also needed to examine and categorize the available visual material. Importantly, this forms the basis of the machine’s learning process.
Discover the innovative power of digital technologies like Machine Learning at drinktec 2022
The advent of Machine Learning makes other innovations for the beverage industry possible. Modern cleaning technologies based on self-learning programs and machine maintenance are already performed using Augmented Reality (AR) data glasses. Machine manufacturers evaluating plant data has also opened up space to use digital services for optimization. As a result, utilities providers have been able to reduce downtimes and limit the use of cleaning solutions and water.
Ready to be amazed by more sparkling ideas and inspiration that translates to efficient and sustainable practices in the beverage industry? You’re in the right place! Drinktec 2022 is the most important convention in the beverage and liquid food industry. Discover the newest technologies and products created with Machine Learning today. Plan your visit to drinktec 2022 from September 12 to 16 in Munich now. Don’t miss out. Check our exhibitor directory to get your ticket.
Want to share your developments and innovations in the beverage industry to an international specialist audience? Then we would very much like to invite you to take part in the next drinktec from September 12 to 16, 2022 in Munich.