TU Bergakademie Freiberg Data Publications

Permanent URI for this collection

http://opara.zih.tu-dresden.de/handle/123456789/21
Data publications from research of Freiberg University of Mining and Technology.

Browse

Browsing TU Bergakademie Freiberg Data Publications by Subject "4::42::403::403-03"

Now showing 1 - 4 of 4
  • Item
    Open Access
    2D Imaging Simulation from 3D Particle Data: Extended Python Notebooks, Simulation Results
    (Technische Universität Bergakademie Freiberg, 2024-07-30) Buchwald, Thomas
    2D projection imaging techniques are simulated with 3D particle data from tomography measurements. This submission contains two distinct parts: Python notebooks to simulate 2D imaging of the particles, and the resulting simulation dataset. This dataset extends a previous one (https://doi.org/10.25532/OPARA-479) by an additional method of particle surface area determination and a larger dataset of randomly oriented projections – ten instead of the previous three. The Python notebooks were created with Jupyter Lab and Anaconda. A environment.yml file is provided that recreates the Anaconda environment. The simulation dataset that results from the provided Python notebooks is made available as CSV files or as pickled Python (pandas) DataFrames. Please refer to the included Readme for a detailed description of the files contained in the archive.
  • Item
    Open Access
    2D Imaging Simulation from 3D Particle Data: Python Notebooks, Particle Datasets, and Simulation Results
    (Technische Universität Bergakademie Freiberg, 2024-05-14) Buchwald, Thomas; Ditscherlein, Ralf
    2D projection imaging techniques are simulated with 3D particle data from tomography measurements. This submission contains three distinct parts: original particle data, Python notebooks to simulate 2D imaging of the particles, and the resulting simulation dataset. The particle data comes as STLs that have been converted from VTK particle data as provided by the PARROT database (https://parrot.tu-freiberg.de/). The STL files are *not* identical with the STLs provided by PARROT as of May 2024! The particle data folder is provided as separate archive because of its large size. The Python notebooks were created with Jupyter Lab and Anaconda. A environment.yml file is provided that recreates the Anaconda environment. The simulation dataset that results from the provided Python notebooks is made available as CSV files or as pickled Python (pandas) DataFrames. Please refer to the included Readme for a detailed description of the files contained in the archive.
  • Item
    Open Access
    Dynamic Image Analysis: Python Notebooks, Particle Datasets, and Simulation Results
    (Technische Universität Bergakademie Freiberg, 2024-08-20) Buchwald, Thomas
    This submission serves as a validation dataset for the simulation results of 2D imaging methods from 3D particle meshes. The submission contains dynamic image analysis data that can be used for validation of simulation results, Python notebooks for extraction of particles and calculation of particle characteristics, and the final results in as pickled Pandas DataFrames and CSV files. This dataset extends to previous submission: https://doi.org/10.25532/OPARA-479 and https://doi.org/10.25532/OPARA-587, which contain the 3D particle dataset and the simulation algorithm notebooks.
  • Item
    Open Access
    Supplementary data of publication: "In-line Image Analysis of Particulate Processes with Deep Learning: Optimizing Training Data Generation via Copy-Paste Augmentation"
    (Technische Universität Bergakademie Freiberg, 2024-04-26) Daus, Sarah
    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.