All PET bottles are marked with an expiration date. Everyone knows that. However, the quality of the marking can vary depending on the speed of the production line and the technology chosen. Conventional image processing software copes well with these markings if the bottle color, speed and bottle content are the same. However, if there are variations, then this software reaches its limits.
However, Deep Learning software has no problem with variations. Thanks to this software, manufacturers can reduce the cost of fine-tuning and maintaining the vision system. This is because the Deep Learning software is trained to automatically adapt to new variations. Thus, the results are completely independent of the bottle color, the marking quality, and the color of the contents.
This software basically has the ability to see and make decisions like a human. This can reduce scrap because there are fewer wrong results.
The OCR tool used is based on the artificial intelligence of the Retina Olive Deep Learning solution. This tool can read more than 50,000 parts/hour with a 99.99% read rate. The developer of the software 3HLE is very confident in this data and guarantees these numbers to the customers.
This demo video shows a typical OCR application in the food and beverage industry. The marking is on transparent, colored and curved PET surfaces.
Conventional OCR tools were able to achieve a read rate of 95% in this application, but only with a single liquid. This result was also achieved only after a few days on which fine adjustments and other tricks were applied. This is because it is important that even poorly readable characters are recognized.
The Retina Olive OCR Tool, on the other hand, already achieved a read rate of 99.99% within a few hours, without any confusion.
You would like to test this tool with your application? Then contact us.