Everyone’s talking about it – but Acrovision will share where the latest AI and Deep Learning advancements in Vision can be put to good use in real-life situations. Acrovision will show that Deep Learning Vision is solving applications not possible with traditional vision but remaining cost-effective

This presentation considers the application of machine learning and artificial intelligence in the rapidly growing field of thermal imaging. Thermal cameras are rapidly becoming an essential tool in the packaging sector – to identify quality issues during sealing and carton closure, but many of these processes cannot be easily defined in a programmatic sense. The processes do not have a clear manufacturing tolerances defining pass or fail criteria and are affected by variations like environment or poor fixturing. Inspections are better defined by learning by examining the characteristics of passed and failed examples, as with machine learning. This presentation considers some real cases and how artificial intelligence can be used to improve inspection performance.

The inspection of parenteral liquids and primary & secondary packaging to ensure patient safety is a critical quality control function in aseptic environments. Pharmacopeia regulations mandate visual inspection of product to safeguard patients. Machine vision is used to achieve a level of resolution, speed and repeatability not possible with manual visual inspection. This presentation details how Crest Solutions automate many inspection and automation functions around vial fill-finish and packaging processes.

Superficially, a label inspection system seems simple. You just point a camera at the label, program in any data and that off you go. Unfortunately, this sort of system isn’t terribly useful and, in all likelihood, will not work at all.The difficulties can be placed in several different categories principally optical, mechanical, print, data, integrity and audit. This presentation will examine each in turn, based on Dimaco’s experience over nearly twenty years of designing such systems. We will highlight the likely technical issues and offer proven solutions to each in turn. Specifically, the presentation will consider the real-world practical problems of running a vision system operated by non-specialist personnel whilst ensuring 100% label verification, minimal false rejects, and highly accurate production data sets.

A lot of our recent projects have been in the food and agricultural sector. With anincreased awareness of reducing throw away during food production vision systemsare a great way to aid towards this goal and work together with automation machinesand robots.Post-harvest vegetable processing is a major contributor to food waste. Improving thisarea of farming is a major challenge for farmers and can not only increase the producequality but also reduce cost.Jana will talk about some of the challenges and solutions faced for this and giveexamples of successfully deployed systems.