Paul Grimm

Paul Grimm

Auto-generated profile, based on publicly available information

Paul Grimm

Faculty of Media, Darmstadt University of Applied Sciences, Darmstadt, Germany

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 May 2026 in Journal on Image and Video Processing
Verified
Journal Article
Open access
PDF file
SPSabrina Pietzsch
Sabrina Pietzsch
JDJan L. Dybeck
Jan L. Dybeck
MRMarkus Roth
Markus Roth
PGPaul Grimm
Paul Grimm
BSBenedikt Schmitz
Benedikt Schmitz
Purpose: We present a comparative study of segmentation methods for high-power laser applications, focusing on two specific challenges: detection of microscopic surface damage on optical components and detection of radiochromic films for reconstructing high-dimensional particle phase space distributions. Methods: Both applications involve complex morphological variations and non-homogeneous contrast conditions, requiring robust and scalable analysis methods. We evaluate two conventional algorithms and two deep learning-based instance segmentation models, including YOLOv8n-seg and a Detectron2-based Mask R-CNN implementation. All models are evaluated on real datasets that reflect the experimental complexities. We focus particular attention to the accuracy of contour detection, using geometric evaluation metrics such as radial contour comparison, Hausdorff distance, Chamfer distance, as well as intersection-over-union, and analysing runtime performance. Results: Our results indicate that the YOLOv8n-seg model outperforms the conventional surface damage segmentation method in accuracy, but with 12 times higher computational requirements. In contrast, for radiochromic films analysing YOLOv8n-seg achieves both higher accuracy and faster evaluation. In comparison to YOLOv8n-seg model, Detectron2-based Mask R-CNN implementation lags in both segmentation performance and runtime. Conclusion: These results highlight the potential of YOLOv8n-seg model in addressing specific data-related challenges in modern laser diagnostics and support their role in the development of next-generation automated analysis systems.

20 December 2021 in International Journal of Semantic Computing
Verified
Journal Article
Closed access
PDF file
MTMarcel Tiator
Marcel Tiator
AKAnna Maria Kerkmann
Anna Maria Kerkmann
CGChristian Geiger
Christian Geiger
PGPaul Grimm
Paul Grimm


9 October 2024
Verified
Conference Paper
Open access
ADAndreas Dietze
Andreas Dietze
YJYvonne Jung
Yvonne Jung
PGPaul Grimm
Paul Grimm
In this arcticle, we present a concept of a tool suite that combines a generic and object-specific difference detection between 3D measurment and 3D planning data (cp. Figure 1). In a Generic Difference Detection (GDD) between measurement and planning data, the objects of interest to be compared are represented by the acquired 3D data (often the solid of a 3D reconstruction) and the 3D planning data in their entirety. An example of this is a shape analysis used as a non-destructive testing method for products from an additive (e.g. 3D printing) or subtractive (e.g. CNC milling) manufacturing process, in which a 3D reconstruction of the produced object is compared with its orignial 3D planning data [2]. The range of applications in this area is very broad and includes quality controls, plagiarism checks and measuring wear and tear. On the other hand, in an Object-Specific Difference Detection (OSDD) the objects of interest are located within the measurement and planning data and are represented by specific components that have to be compared. Here, the sector of digital construction monitoring can be mentioned as an example, in which the construction process and associated specific construction components, such as walls, passages and window openings, are checked against the planning data based on acquired measurement data [1]. Both difference detection techniques work, regardless of whether the object of interest is present in both the measurement and planning data or only in the measurement or planning data (e.g. after a building refurbishment). Besides the benefit of quality assurance following a production or construction process that is covered by both difference detection concepts, an early detection of errors is of great advantage. For example, follow-up costs after a construction process (e.g. building) can be reduced or avoided by identifying errors during the construction process. In addition, this can also help to ensure that the schedule is adhered to. Depending on the application, there is also the option of feeding identified deviations or errors back into the planning data in order to achieve synchronization between the planning data and the actual state of the object, in case the differences are intentional. While our approach for GDD is limited to a shape similarity analysis of two 3D objects and a subsequent real-time visualization of detected differences, our method for OSDD can already feed detected deviations and errors back into the 3D planning data and allows a collaborative result visualization based on a multi-codal presentation (e.g. tabular data or 3D rendering).

14 August 2026
Verified
Book Chapter
Closed access
PDF file
RDRalf Dörner
Ralf Dörner
WBWolfgang Broll
Wolfgang Broll
PGPaul Grimm
Paul Grimm
BJBernhard Jung
Bernhard Jung