Högskolan i Skövde

Defect detection in sealing processes is so necessary for guaranteeing product quality and safety across several industries. Traditional defect detection techniques frequently rely on subjective and manual inspection methods that leads to limitations in accuracy and efficiency. This project focusses mainly on these challenges by developing an automated defect detection system using machine learning techniques. 

The primary objective is to improve upon the basic standards that are already established by traditional methods that helps in enhancing the effectiveness and accuracy of defect identification in sealing processes. The proposed system utilises deep learning algorithms, including convolutional neural networks (CNNs) and other advanced architectures available as of now, to detect sealing defects such as cracks, uneven surfaces, and improper seals. 

The methodology of the project consists of the following key steps that includes data collection, pre-processing, model training, validation, and testing. Performance metrics such as detection accuracy, false positive rate, and processing time are also used to effectively assess the productivity of the automated system. 

The results exhibit significant improvements over the basic standard, with the automated system attaining higher accuracy and efficiency in defect detection compared to traditional methods. Statistical analysis endorses the practical significance of these improvements, focussing on the potential of machine learning techniques to increase the quality of defect detection in sealing processes. 

Overall, this project donates to the advancement of industrial sealing techniques, leading to improvements in product quality and safety. By providing a general overview of the project's background, methodology, and outcomes, this report aims to highlight the impact of using machine learning for defect detection in sealing processes.