FEATURE Machine vision
Image courtesy of Asspe
HALCON machine vision software from MVTec is providing quality assurance that is as good as a human examiner – or even better, at pharmaceutical company Aspen Notre-Dame-De-Bondeville
T
hanks to ongoing technological advances, more and more tasks can be automated with machine vision. A particularly challenging project was successfully completed at the pharmaceutical company Aspen Notre-Dame-De-Bondeville, for foreign objects. With the HALCON machine vision software from MVTec and the support of MVTec’s customer service team, quality assurance was improved.
increasingly common in industrial settings and makes it possible to automate even more Machine vision is particularly well suited to quickly available as training data in production environments.
Aspen, headquartered in Durban, South from the advantages of machine vision in combination with deep learning. At the company’s Notre-Dame-de-Bondeville in France, the company weighs and mixes the them into ampoules in a subsequent process. “Our goal was to automate the inspection of ampoules for possible foreign particles. Quality assurance of pharmaceutical products is extremely important. Therefore, it was essential that the new solution matched the detection rates of the previous process, which involved inspection by human operators, or ideally even surpassed them,” explained Mickael Denis, Manager Operationnel Vision Industrielle. Vincent Trombetta, Automatic Visual Inspection Expert at Aspen, continued: “It was clear that such a task could only be
26 January 2026 | Automation
automated using deep learning technologies. For the implementation, which required a great deal of expertise, we relied on the consulting services of MVTec Software. Since the machine vision solution on the inspection machine had already been implemented with MVTec HALCON, it made sense to use the services directly from the manufacturer.”
carried out in one machine, this process is hygienic. Foreign bodies can hardly get into the ampoules. Nevertheless, this process takes place under clean room conditions to further minimise the risk of contamination. sealed, they are transported to an inspection and packaging area. The ampoules and their contents are then checked for defects. Previously, this process involved employees picking up each ampoule individually and checking it from all sides to see if the ampoule foreign objects in the liquid. The big challenge here is that the contents of the ampoules to distinguish from foreign objects. The easy to detect, even for inspectors, making manual inspection very time-consuming and costly. For this reason, the new process was automated. Denis explained: “Since the inspection has to be done visually, it was clear that we could only implement the process with machine vision and no other technology. We also had to adapt to the particularly demanding validation processes that apply in the pharmaceutical industry. This ensures
p n NDB DEEP LEARNING OFFERS SMARTER INSPECTION DEEP LEARNING OFFERS SMARTER INSPECTION
that the new system tests the ampoules at the required speed, but at the same time with the highest accuracy and robustness.” A total of 12 cameras that comply with the industry standard GigE Vision are used in the new system. Good lighting is also important to make the particles clearly visible. Machine Vision is performed on an industrial PC, using MVTec HALCON. Due to the extremely deep learning had clear advantages over classic rule-based methods. With classic machine detect the defects.
In practice, the test vials are manually placed on the conveyor belt of the system and thus reach the inspection machine. The 12 cameras are positioned so that they capture a total of up stations. The large number of images is helpful for deep learning-based inspection, as there are images in which the particles are not visible angle. It should also be noted that a particle is number of images. This successfully reduces the number of false positives. Once the images have been captured, they are transmitted to MVTec HALCON. There, various machine vision methods are used to perform the checks. Aspen uses the deep learning-based semantic segmentation included in HALCON to detect foreign matters. In addition to inspecting the liquid for particles, other inspection tasks are also performed in parallel with HALCON. These include so-called cosmetic defects. correct, whether the colour is appropriate, and whether the closure complies with such as matching and blob analysis are used for these tasks. At the end of the inspection, a decision is made as to whether the ampoule is OK or NOK.
post-processing, and recompiling the data previously labelled by Aspen, and then training it multiple times. “With the support of our colleagues at
increasing the error detection rate and reducing false negative results,” explained Trombetta.
MVTec
www.mvtec.com/products/halcon
automationmagazine.co.uk
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40
