This dataset contains the supplementary data of the open-access publication "In-line Image Analysis of Particulate Processes with Deep Learning: Optimizing Training Data Generation via Copy-Paste Augmentation". It provides all the necessary data and scripts to replicate the results of the study and to further use the synthetic data generation to train one's own deep learning image segmentation model. The original scripts have been slightly modified to make them more suitable for general use, e.g. parameters have been passed to an argument parser instead of hard-coding them whenever possible.

Abstract of Paper: Monitoring particle properties directly in the process using in-situ microscopy can provide valuable input to control loops, improve process understanding and facilitate process optimization. However, obtaining reliable information from these images remains a challenge, especially for higher solids concentrations or agglomerating systems. Recent studies have successfully applied deep learning models to extract particle characteristics from in-situ image data. Despite these advances, the problem of generating training data has not been properly addressed. Manual annotation is time-consuming and prone to bias due to high particle counts, particle overlap, and out-of-focus objects. This paper presents a new approach to generate training data for segmentation models by combining conventional segmentation methods with copy-paste augmentation. A case study was conducted in which depth filtration experiments were performed on irregularly shaped alumina particles. An instance segmentation model trained on data generated using the proposed approach successfully detected and characterized particles, even at high solids concentrations.