Quality assessment in industrial applications is often carried out through visual inspection, usually performed or supported by human domain experts. However, the manual visual inspection of processes and products is error-prone and expensive. It is therefore not surprising that the automation of visual inspection in manufacturing and maintenance is heavily researched and discussed. The use of artificial intelligence as an approach to visual inspection in industrial applications has been considered for decades. Recent successes, driven by advances in deep learning, present a possible paradigm shift and have the potential to facilitate automated visual inspection, even under complex environmental conditions. For this reason, we explore the question of to what extent deep learning is already being used in the field of automated visual inspection and which potential improvements to the state of the art could be realized utilizing concepts from academic research. By conducting an extensive review of the openly accessible literature, we provide an overview of proposed and in-use deep-learning models presented in recent years. Our survey consists of 196 open-access publications, of which 31.7% are manufacturing use cases and 68.3% are maintenance use cases. Furthermore, the survey also shows that the majority of the models currently in use are based on convolutional neural networks, the current de facto standard for image classification, object recognition, or object segmentation tasks. Nevertheless, we see the emergence of vision transformer models that seem to outperform convolutional neural networks but require more resources, which also opens up new research opportunities for the future. Another finding is that in 97% of the publications, the authors use supervised learning techniques to train their models. However, with the median dataset size consisting of 2500 samples, deep-learning models cannot be trained from scratch, so it would be beneficial to use other training paradigms, such as self-supervised learning. In addition, we identified a gap of approximately three years between approaches from deep-learning-based computer vision being published and their introduction in industrial visual inspection applications. Based on our findings, we additionally discuss potential future developments in the area of automated visual inspection.

Artificial intelligence as an approach to visual inspection in industrial applications has been considered for decades. Recent successes, driven by advances in deep learning, present a potential paradigm shift and have the potential to facilitate an automated visual inspection, even under complex environmental conditions. Thereby, convolutional neural networks (CNN) have been the de facto standard in deep-learning-based computer vision (CV) for the last 10 years. Recently, attention-based vision transformer architectures emerged and surpassed the performance of CNNs on benchmark datasets, regarding regular CV tasks, such as image classification, object detection, or segmentation. Nevertheless, despite their outstanding results, the application of vision transformers to real world visual inspection is sparse. We suspect that this is likely due to the assumption that they require enormous amounts of data to be effective. In this study, we evaluate this assumption. For this, we perform a systematic comparison of seven widely-used state-of-the-art CNN and transformer based architectures trained in three different use cases in the domain of visual damage assessment for railway freight car maintenance. We show that vision transformer models achieve at least equivalent performance to CNNs in industrial applications with sparse data available, and significantly surpass them in increasingly complex tasks.

The application of neural networks in production scenarios has been growing over the last years. However, their innate black-box character and the lack of transparency and interpretability of their decision making hinders their application in safety critical scenarios and scenarios with high failure costs. We present an approach to facilitate transparency of failure predictions performed by convolutional neural networks processing sensory time series data from a deep drawing manufacturing process for predictive quality purposes. Our approach enables domain experts to comprehend why a failure is predicted, enabling them to better assess whether to intervene into the production process or not.

In recent years, Explainable AI has gained traction as an approach to enhancing model interpretability and transparency, particularly in complex models such as detection transformers. Despite rapid advancements, a substantial research gap remains in understanding the distinct roles of internal components - knowledge that is essential for improving transparency and efficiency. Inspired by neuroscientific ablation studies, which investigate the functions of brain regions through selective impairment, we systematically analyze the impact of ablating key components in three state-of-the-art detection transformer models: Detection transformer (DETR), deformable detection transformer (DDETR), and DETR with improved denoising anchor boxes (DINO). The ablations target query embeddings, encoder and decoder multi-head self-attentions (MHSA) as well as decoder multi-head cross-attention (MHCA) layers. We evaluate the effects of these ablations on the performance metrics gIoU and F1-score, quantifying effects on both the classification and regression sub-tasks on the COCO dataset. To facilitate reproducibility and future research, we publicly release the DeepDissect library. Our findings reveal model-specific resilience patterns: while DETR is particularly sensitive to ablations in encoder MHSA and decoder MHCA, DDETR's multi-scale deformable attention enhances robustness, and DINO exhibits the greatest resilience due to its look-forward twice update rule, which helps distributing knowledge across blocks. These insights also expose structural redundancies, particularly in DDETR's and DINO's decoder MHCA layers, highlighting opportunities for model simplification without sacrificing performance. This study advances XAI for DETRs by clarifying the contributions of internal components to model performance, offering insights to optimize and improve transparency and efficiency in critical applications.