|
|
Automatic Image Analysis for Production Quality Control of Textile
Sýkorová, Tereza ; Dobeš, Petr (referee) ; Zemčík, Pavel (advisor)
This work deals with the classification of defects that occur in the production of nonwovens. The defect classification task is part of a system for automatic production quality control. The goal is to implement a method that will classify problematic defect classes with sufficient accuracy. That was achieved using convolutional neural networks (CNN). The best results were achieved by the EfficientNet network, which had an accuracy of 81% when evaluated by cross-validation on an available dataset. Within the work, a number of experiments are performed, which are focused on the modification of input data. The influence of the shape and composition of the input images on the final classification is examined. A CNN model was also implemented, which uses additional information for classification in addition to the image.
|
|
|
Classification of board defects in semiconductor manufacturing
Jašek, Filip ; Vágner, Martin (referee) ; Dřínovský, Jiří (advisor)
This diploma thesis focuses on detecting defects in semiconductor wafer manufacturing. It explores methods for identifying faulty chips and controlling yield during production. To classify defects machine learning techniques are used. Initially, ResNet18 architecture was used for inference, but low accuracy was attributed to limited input data. Transfer learning with ResNet50v2 was then attempted, resulting in improved metric with different dataset. Hyperparameter tuning and data augmentations were also explored. The study found that autoencoders for data compression during inference increased speed but led to degraded evaluation metrics.
|
|
|
Classification of board defects in semiconductor manufacturing
Jašek, Filip ; Vágner, Martin (referee) ; Dřínovský, Jiří (advisor)
This diploma thesis focuses on detecting defects in semiconductor wafer manufacturing. It explores methods for identifying faulty chips and controlling yield during production. To classify defects machine learning techniques are used. Initially, ResNet18 architecture was used for inference, but low accuracy was attributed to limited input data. Transfer learning with ResNet50v2 was then attempted, resulting in improved metric with different dataset. Hyperparameter tuning and data augmentations were also explored. The study found that autoencoders for data compression during inference increased speed but led to degraded evaluation metrics.
|
|
|
Automatic Image Analysis for Production Quality Control of Textile
Sýkorová, Tereza ; Dobeš, Petr (referee) ; Zemčík, Pavel (advisor)
This work deals with the classification of defects that occur in the production of nonwovens. The defect classification task is part of a system for automatic production quality control. The goal is to implement a method that will classify problematic defect classes with sufficient accuracy. That was achieved using convolutional neural networks (CNN). The best results were achieved by the EfficientNet network, which had an accuracy of 81% when evaluated by cross-validation on an available dataset. Within the work, a number of experiments are performed, which are focused on the modification of input data. The influence of the shape and composition of the input images on the final classification is examined. A CNN model was also implemented, which uses additional information for classification in addition to the image.
|
guest ::
