Classification of gravure printed patterns using convolutional neural networks (Python code)

This dataset contains Python code ('code_DeepLearn_ImgClass.zip') for automated classification of gravure printed patterns from the [HYPA-p](https://tudatalib.ulb.tu-darmstadt.de/handle/tudatalib/3841) dataset. The developed algorithm performs supervised deep learning of convolutional neural networks (CNNs) on labeled data ('CNN_dataset.zip'), i.e. selected, labeled 'S-subfields' from the HYPA-p dataset. 'CNN_dataset.zip' is a subset from the images in the folder 'labeled_data.zip', which can be created with the provided Python code. PyTorch is used as a deep learning framework. The Python code yields trained CNNs, which can be used for automated classification of unlabeled data from the HYPA-p dataset. Well-known, pre-trained network architectures like Densenet-161 or MobileNetV2 are used as a starting point for training. Several trained CNNs are included in this submission, see 'trained_CNN_models.zip'. Further information can be found in the dissertation of Pauline Rothmann-Brumm (2023) and in the provided README-file.