We have developed an automatic quality control system for consumer product labels such as detergents, cleaners, shampoos, make-up or adhesives.
Before our solution, the quality control process was done manually. During this process, each of the characteristics of the labels (dimensions, shape, colors, brand, barcode, ingredients, contact information, safety pictograms, recycling pictograms,…) were compared with the data stored in the system. Now, this comparision is done by an automatic AI-based process with state-of-the-art technology.
The manual process was a tedious, time-consuming and error-prone process. Thanks to AI, the process is automated, saving time, improving productivity, increasing the accuracy of the process and allowing our client and its employees to dedicate more time to more important tasks.
For the development of this project, we’ve used Natural Language Processing and Computer Vision technology. Label specifications have a structured textual component, an unstructured textual component and a visual component. Thanks to NLP, we can understand the text that is distributed on the label, such as instructions for use, safety warnings or ingredients, and certify that they are correct. Using Deep Learning Object Detection technology, we can recognize product brand logos, as well as recycling, safety, information and other pictograms.
Using segmentation neural networks, we detect the precise edges of the label, after spatial calibration, to check if the shape and dimensions of the label are as required by the production department. Other computer vision technologies, allow us to check that the printing colors are correct and the label dimensions are calibrated, among other things.
(1) Source: URL [www.henkel.com/company]
Henkel logo © 2022 Henkel AG & Co. KGaA. All rights reserved.