Generating surface topography of paperboard using diffusion models
(2025) In Master's Theses in Mathematical Sciences FMSM01 20242Mathematical Statistics
- Abstract
- This thesis uses the unsupervised deep learning method of generative diffusion models to generate images of paperboard surfaces. As the images are intended to be
used as input to various simulations by food packaging company Tetra Pak, simply
looking at them is not sufficient to determine their quality. For this reason, some
new evaluation metrics, specific to paperboard images, are proposed and used. We
present model architectures and training parameters that allow for sampling of new
images, examples of generated images and the evaluation of them. The generated
images bear a close resemblance to real images both when visually inspected and
in terms of the used evaluation metrics. Finally, an alternative use of the models
in this... (More) - This thesis uses the unsupervised deep learning method of generative diffusion models to generate images of paperboard surfaces. As the images are intended to be
used as input to various simulations by food packaging company Tetra Pak, simply
looking at them is not sufficient to determine their quality. For this reason, some
new evaluation metrics, specific to paperboard images, are proposed and used. We
present model architectures and training parameters that allow for sampling of new
images, examples of generated images and the evaluation of them. The generated
images bear a close resemblance to real images both when visually inspected and
in terms of the used evaluation metrics. Finally, an alternative use of the models
in this project is explained. It is possible to use a model to generate images larger than what it was trained on. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9184892
- author
- Loman, Ale LU
- supervisor
- organization
- course
- FMSM01 20242
- year
- 2025
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Generative Diffusion Models, Diffusion Models, Generative AI, Paperboard Topography
- publication/series
- Master's Theses in Mathematical Sciences
- report number
- LUTFMS-3510-2025
- ISSN
- 1404-6342
- other publication id
- 2025:E5
- language
- English
- id
- 9184892
- date added to LUP
- 2025-02-17 15:49:14
- date last changed
- 2025-02-26 14:37:35
@misc{9184892, abstract = {{This thesis uses the unsupervised deep learning method of generative diffusion models to generate images of paperboard surfaces. As the images are intended to be used as input to various simulations by food packaging company Tetra Pak, simply looking at them is not sufficient to determine their quality. For this reason, some new evaluation metrics, specific to paperboard images, are proposed and used. We present model architectures and training parameters that allow for sampling of new images, examples of generated images and the evaluation of them. The generated images bear a close resemblance to real images both when visually inspected and in terms of the used evaluation metrics. Finally, an alternative use of the models in this project is explained. It is possible to use a model to generate images larger than what it was trained on.}}, author = {{Loman, Ale}}, issn = {{1404-6342}}, language = {{eng}}, note = {{Student Paper}}, series = {{Master's Theses in Mathematical Sciences}}, title = {{Generating surface topography of paperboard using diffusion models}}, year = {{2025}}, }