This paper provides a comprehensive survey of open-access literature on the application of deep learning (DL) in automated visual inspection (AVI) for manufacturing and maintenance. The authors review 196 publications, with 31.7% focusing on manufacturing and 68.3% on maintenance. Convolutional neural networks (CNNs) are the most prevalent model type, though vision transformer models are emerging and showing promise. Supervised learning is the dominant training paradigm, but the median dataset size of 2500 samples suggests the need for alternative paradigms like self-supervised learning. The survey highlights a gap of about three years between the publication and industrial application of deep-learning-based computer vision techniques. The paper also discusses the requirements for DL-based AVI, categorizes use cases, and evaluates the performance of models across different tasks. It concludes with an analysis of data characteristics and performance metrics, identifying areas for future research and potential improvements in AVI.This paper provides a comprehensive survey of open-access literature on the application of deep learning (DL) in automated visual inspection (AVI) for manufacturing and maintenance. The authors review 196 publications, with 31.7% focusing on manufacturing and 68.3% on maintenance. Convolutional neural networks (CNNs) are the most prevalent model type, though vision transformer models are emerging and showing promise. Supervised learning is the dominant training paradigm, but the median dataset size of 2500 samples suggests the need for alternative paradigms like self-supervised learning. The survey highlights a gap of about three years between the publication and industrial application of deep-learning-based computer vision techniques. The paper also discusses the requirements for DL-based AVI, categorizes use cases, and evaluates the performance of models across different tasks. It concludes with an analysis of data characteristics and performance metrics, identifying areas for future research and potential improvements in AVI.