One of the most promising use-cases for machine learning in industrial manufacturing is the early detection of defective products using a quality control system. Such a system can save costs and reduces human errors due to the monotonous nature of visual inspections. Today, a rich body of research exists which employs machine learning methods to identify rare defective products in unbalanced visual quality control datasets. These methods typically rely on two components: A visual backbone to capture the features of the input image and an anomaly detection algorithm that decides if these features are within an expected distribution. With the rise of transformer architecture as visual backbones of choice, there exists now a great variety of different combinations of these two components, ranging all along the trade-off between detection quality and inference time. Facing this variety, practitioners in the field often have to spend a considerable amount of time on researching the right combination for their use-case at hand. Our contribution is to help practitioners with this choice by reviewing and evaluating current vision transformer models together with anomaly detection methods. For this, we chose SotA models of both disciplines, combined them and evaluated them towards the goal of having small, fast and efficient anomaly detection models suitable for industrial manufacturing. We evaluated the results of our experiments on the well-known MVTecAD and BTAD datasets. Moreover, we give guidelines for choosing a suitable model architecture for a quality control system in practice, considering given use-case and hardware constraints.

In industrial manufacturing of glass bottles, quality control of bottle prints is necessary as numerous factors can negatively affect the printing process. Even minor defects in the bottle prints must be detected despite reflections in the glass or manufacturing-related deviations. In cooperation with our medium-sized industrial partner, two ML-based approaches for quality control of these bottle prints were developed and evaluated, which can also be used in this challenging scenario. Our first approach utilized different filters to supress reflections (e.g. Sobel or Canny) and image quality metrics for image comparison (e.g. MSE or SSIM) as features for different supervised classification models (e.g. SVM or k-Neighbors), which resulted in an accuracy of 84%. The images were aligned based on the ORB algorithm, which allowed us to estimate the rotations of the prints, which may serve as an indicator for anomalies in the manufacturing process. In our second approach, we fine-tuned different pre-trained CNN models (e.g. ResNet or VGG) for binary classification, which resulted in an accuracy of 87%. Utilizing Grad-Cam on our fine-tuned ResNet-34, we were able to localize and visualize frequently defective bottle print regions. This method allowed us to provide insights that could be used to optimize the actual manufacturing process. This paper also describes our general approach and the challenges we encountered in practice with data collection during ongoing production, unsupervised preselection, and labeling.

Conferences, media, vendors, automation industry consultants, business consultants, and even politicians are discussing and making presentations about how the Internet of Things (IoT) and Industry 4.0 are creating a revolution in manufacturing. I am convinced we are at a juncture of major industrial automation changes driven by technology advancements. The digital revolution of business functions, including accounting, supply chain, human resources, procurement, customer services, business intelligence, and distribution management, has been refined over multiple generations. In contrast, the industrial and process automation industries have not transformed at the same rate. They must be digitized now for manufacturers to compete. At the end of this article I have the results of a small survey of readers that may be interesting.

In the past few years, object detection has attracted a lot of attention in the context of human-robot collaboration and Industry 5.0 due to enormous quality improvements in deep learning technologies. In many applications, object detection models have to be able to quickly adapt to a changing environment, i.e., to learn new objects. A crucial but challenging prerequisite for this is the automatic generation of new training data which currently still limits the broad application of object detection methods in industrial manufacturing. In this work, we discuss how to adapt state-of-the-art object detection methods for the task of automatic bounding box annotation for the use case where the background is homogeneous and the object's label is provided by a human. We compare an adapted version of Faster R-CNN and the Scaled Yolov4-p5 architecture and show that both can be trained to distinguish unknown objects from a complex but homogeneous background using only a small amount of training data.

Importantly, they can also prevent costly defects and avoid operational inefficiencies. While COVID-19 sped up the pace of adoption for many industries, including industrial manufacturing, manufacturing companies have historically embraced new ways of working. Manufacturers were early endorsers of Kaizen, Six Sigma, and Lean, known business improvement models with direct impacts to the continuous improvement methodology critical to manufacturing processes. And now, AI is being embraced for its ability to make supply chains more flexible -- mostly to evaluate vulnerabilities identified during the COVID-19 pandemic among their suppliers and in the supply chain itself -- reduce costs, and fully leverage human talent and intelligence. According to a new KPMG report, Thriving in an AI World, 93% of industrial manufacturing respondents indicated they have moderate or fully functional AI, primarily machine learning technologies, implemented into their processes.

In a recent survey, "2021 Thriving in an AI World," KPMG found that across every industry--manufacturing to technology to retail--the adoption of artificial intelligence (AI) is increasing year over year. Part of the reason is digital transformation is moving faster, which helps companies start to move exponentially faster. But, as Cliff Justice, US leader for enterprise innovation at KPMG posits, "Covid-19 has accelerated the pace of digital in many ways, across many types of technologies." Justice continues, "This is where we are starting to experience such a rapid pace of exponential change that it's very difficult for most people to understand the progress." But understand it they must because "artificial intelligence is evolving at a very rapid pace." Justice challenges us to think about AI in a different way, "more like a relationship with technology, as opposed to a tool that we program," because he says, "AI is something that evolves and learns and develops the more it gets exposed to humans." If your business is a laggard in AI adoption, Justice has some cautious encouragement, "[the] AI-centric world is going to accelerate everything digital has to offer." Business Lab is hosted by Laurel Ruma, editorial director of Insights, the custom publishing division of MIT Technology Review.

Specifically, 55 percent of business leaders in industrial manufacturing and 49 percent in retail and tech told KPMG that "AI is moving faster than it should in their industry." Moreover, these concerns are exceedingly pronounced within the section of leaders from small companies at 63 percent, leaders with high levels of AI knowledge at 51 percent and by Gen Z and Millennial business leaders at 51 percent. Notably, many business leaders with interest in AI implementation said COVID-19 has influenced their company's plans to adopt the technology and contributed to a quicker pace of adoption, including 53 percent from retail, 57 percent from tech, 72 percent from industrial manufacturing, and 37 percent from healthcare and life sciences. In fact, according to the survey 93 percent of leaders in industrial manufacturing, 84 percent in financial services, 83 percent in tech, 81 percent in retail, 77 percent in life sciences, 67 percent in health care and 61 percent in government, said AI is at least "moderately functional" in their organizations. Though 78 percent of retail business leaders said it is difficult to stay on top of the constantly evolving AI landscape -- expressing the sentiment more than leaders in other categories. Looking ahead, 53 percent of retail business leaders said they expect AI to have the greatest impact on customer intelligence, while 50 percent stated inventory management and 49 percent cited chatbots for customer service.

According to a new report by KPMG, "Thriving in an AI World," industry leaders may be experiencing an effect that the company is calling "COVID-19 whiplash" after a year of highly accelerated technology adoption. Specifically, 55 percent of business leaders in industrial manufacturing and 49 percent in retail and tech told KPMG that "AI is moving faster than it should in their industry." Moreover, these concerns are exceedingly pronounced within the section of leaders from small companies at 63 percent, leaders with high levels of AI knowledge at 51 percent and by Gen Z and Millennial business leaders at 51 percent. Notably, many business leaders with interest in AI implementation said COVID-19 has influenced their company's plans to adopt the technology and contributed to a quicker pace of adoption, including 53 percent from retail, 57 percent from tech, 72 percent from industrial manufacturing, and 37 percent from healthcare and life sciences. "Just one year ago, [our report] signaled on all accounts that AI was starting to have real impact across industries, yet industry leaders told us that they felt it was not being implemented fast enough," said Traci Gusher, principal of artificial intelligence at KPMG. "Fast forward to today, industry leaders are telling us they are experiencing what we at KPMG are calling COVID-19 whiplash, with AI adoption literally skyrocketing as a result of the pandemic. Now, Industry leaders are saying it's moving too fast."

Real-world AI deployments surged over the past year as companies sought to remain competitive during the coronavirus pandemic, according to a new study released today by KPMG. However, even as they expanded and accelerated their AI projects, organizations expressed concerns about ethics and bias, and suggested AI might be getting ahead of regulations. KPMG's study, called "Thriving in an AI World," replicates a study conducted before COVID-19 upended our world a year ago. That provided KPMG Principal of AI, Traci Gusher, a convenient baseline to test how AI deployments have been impacted by COVID-19. "Over half the business leaders that we talked to said that AI is at least moderately to fully functional in their organization, which is a significant increase," Gusher says.

A survey published by KPMG today suggests that a large number of organizations have increased their investments in AI during the pandemic to the point that executives are now concerned about moving too fast. In fact, most of the survey respondents cited a definite need for increased AI regulation. The survey covered 950 business decision-makers and/or IT decision-makers with at least a moderate amount of AI knowledge at companies with more than $1 billion in revenue. It finds AI technologies are most likely to be moderately to fully employed in industrial manufacturing (93%), financial services (84%), technology (83%), retail (81%), life sciences (77%), health care (67%), and government (61%) sectors. Survey respondents all cited the pandemic as a factor that drove increased adoption of AI in the last year, though the degree varied by sector from industrial manufacturing (72%) to technology (57%), retail (53%), government (44%), financial services (42%), and health care and life sciences (37%). Many respondents also noted that AI technology is moving too fast for their comfort in industrial manufacturing (55%), technology (49%), retail (49%), life sciences (47%), financial services (37%), government (37%), and health care (35%) sectors.