F&B manufacturers everywhere constantly deal with two top challenges – how to meet and exceed production targets, and ensure high quality control at the same time?
Especially in production lines that deal with natural products which are different from one another – in size, shape and colour – how can precise measurements be speedily carried out without having to program lots of parameters that have to be manually fine-tuned?
A leading manufacturer of frozen dough and bakery products in Spain had a classic case of these issues. In their production lines, they had to achieve 100% quality control, so all visual parameters of their product in different phases of production had to be controlled.
Adding to their challenges, there were many difficulties regarding the identification and classification of the products as the natural and handmade products were all different from each other, and the production lines were also subject to subjective parameters such as the degree of toasting.
To solve these challenges, the answer was found in none other than – deep learning in machine vision.
Deep learning solutions can classify products twice as fast, with significantly reduced error rate, without a very high programming effort. Furthermore, it can control all quality parameters with 100% analysis of the production.
DEEP LEARNING IN BREAD PRODUCTION
For the Spanish bakery products manufacturer, MVTec HALCON was adopted to provide a complete machine vision solution as its highly robust software allows users to combine ‘traditional’ image analysis with the most advanced deep learning algorithms.
In this, a solution was developed with substantial use of HALCON libraries, together with an interface that was tailored to the customer’s needs. High-resolution colour line scan cameras and specific lighting, as well as 3D metrology systems were further added to facilitate the analysis of very subtle colour and texture parameters
With this, 2D and 3D metrology can be performed alongside deep-learning-based segmentation and classification for:
- Identification of the baked bread on the production line
- Measurements on each product, depending on product type
- Classification of the product according to visual parameters
The result – up to 140 pieces can be analysed per minute, 24 hours a day, and non-conforming products are automatically sorted out, resulting in an increase in production by 30%!
And sweetening the deal further:
- High programming effort was not involved because of the integrated deep learning functions
- No maintenance is required of the system
IS DEEP LEARNING POSSIBLE IN OTHER PRODUCTION LINES?
Today, Deep Learning technologies are enabling an increasing number of applications in almost every industry that were not possible before, and even significantly improving the performance of existing applications.
In particular, in the field of data classification, deep learning is benefitting factory automation and production lines that use machine vision, especially when parameters cannot be clearly defined.
As a result, product identification, sorting, measurement, visual quality checks and inspection can be executed at extremely fast speed, with high accuracy
To top it all, the Halcon Deep Learning tool by MVTec works with most vision cameras, and can extend the power of traditional machine vision in the Food and Beverage industry, as well as in package and logistics, automotive, pharmaceutical, and semiconductor production.
Wonder how this is done?
Come, discover how classification and character recognition challenges can be effectively solved with the power of Halcon Deep Learning at the LINXDays Southeast Asia 2021 Webinar.
All Webinar participants are entitled to an exclusive ‘Halcon Orientation Training’ worth $800, which comes with a 1-month trial of Halcon and immediate hands-on training.
Sign Up Now!