Thomas Rösgen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Thomas Rösgen has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Claudio Mucignat
Laboratory for Computational Engineering, Swiss Federal Laboratories for Materials Science and Technology, Dübendorf, Switzerland
4 shared publications - Ivan Lunati
Laboratory for Computational Engineering, Swiss Federal Laboratories for Materials Science and Technology, Dübendorf, Switzerland
4 shared publications - Markus J. Schmidt
Institute of Fluid Dynamics, ETH Zurich, 8092 Zurich, Switzerland
4 shared publications - Alexander Rusch
Scientific Computing in Computer Science, Department of Informatics, Technical University of Munich, Garching, 85748, Germany
2 shared publications - Jiggar Shah
Laboratory for Multiscale Studies in Building Physics, Swiss Federal Laboratories for Materials Science and Technology, Dübendorf, Switzerland
2 shared publications - Lento Manickathan
Laboratory for Multiscale Studies in Building Physics, Swiss Federal Laboratories for Materials Science and Technology, Dübendorf, Switzerland
2 shared publications
Latest Publications
- Open Access
- 16 December 2025
- Published by Springer Nature in Experiments in Fluids
- Vol. 67 (1) , 7
Explicit illumination correction for error reduction in two-dye LIF thermometry
We propose a method to perform accurate temperature measurements using laser-induced fluorescence (LIF). We use a sCMOS color camera and a two-dye solution consisting of RuPhen and fluorescein, excited at 450 nm . By varying the relative concentration of the dyes, we can tune the temperature sensitivity of the color channels. This enables a robust laser power correction, reducing the effects of experimental noise compared to the convectional ratiometric approach. Furthermore, the overall temperature sensitivity is only slightly lower compared to that of the temperature-sensitive dye, which is not the case if a ratiometric analysis is performed. We demonstrate the capabilities of our method using a benchtop setup with precisely controlled temperatures. Our error analysis shows that an accuracy better than 0.5 $$^\circ C$$ can be achieved. The correction method can be applied to other fluorescence measurement techniques, including pressure-sensitive paints (PSPs).- 16 December 2025
- Published by Springer Nature in Experiments in Fluids
- Vol. 67 (1) , 7
Open AccessExplicit illumination correction for error reduction in two-dye LIF thermometry
We propose a method to perform accurate temperature measurements using laser-induced fluorescence (LIF). We use a sCMOS color camera and a two-dye solution consisting of RuPhen and fluorescein, excited at 450 nm . By varying the relative concentration of the dyes, we can tune the temperature sensitivity of the color channels. This enables a robust laser power correction, reducing the effects of experimental noise compared to the convectional ratiometric approach. Furthermore, the overall temperature sensitivity is only slightly lower compared to that of the temperature-sensitive dye, which is not the case if a ratiometric analysis is performed. We demonstrate the capabilities of our method using a benchtop setup with precisely controlled temperatures. Our error analysis shows that an accuracy better than 0.5 $$^\circ C$$ can be achieved. The correction method can be applied to other fluorescence measurement techniques, including pressure-sensitive paints (PSPs). - Open Access
- 17 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-2
Unveiling the Invisible: Interactive Spatial Sensing Transforms Air Flow Measurement
This engaging experience transforms participants into aerodynamicists, enabling them to explore air flow fields firsthand. Utilizing groundbreaking augmented reality visualizations alongside an active learning measurement system, individuals wield a sensor with inside-out tracking, allowing them to navigate and directly uncover the dynamics of air flow. The system generates real-time visualizations of measured air currents, making the usually complex task of evaluating flow fields more accessible. Assisted by a companion artificial intelligence, users are directed to subsequent measurement points, effectively integrating them into the algorithm’s process. This innovative approach not only simplifies the examination of air flow fields but also enriches the investigative experience by making participants active contributors to the exploration.
- 17 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-2
Open AccessUnveiling the Invisible: Interactive Spatial Sensing Transforms Air Flow Measurement
This engaging experience transforms participants into aerodynamicists, enabling them to explore air flow fields firsthand. Utilizing groundbreaking augmented reality visualizations alongside an active learning measurement system, individuals wield a sensor with inside-out tracking, allowing them to navigate and directly uncover the dynamics of air flow. The system generates real-time visualizations of measured air currents, making the usually complex task of evaluating flow fields more accessible. Assisted by a companion artificial intelligence, users are directed to subsequent measurement points, effectively integrating them into the algorithm’s process. This innovative approach not only simplifies the examination of air flow fields but also enriches the investigative experience by making participants active contributors to the exploration.
- Open Access
- 19 February 2024
- Published by Cambridge University Press (CUP) in Flow
- Vol. 4, E2
Luminescence anisotropy and vorticity magnitude of a free turbulent jet
In the field of experimental fluid dynamics, the direct measurement of vorticity remains a challenge, even though it plays a crucial role in understanding turbulent flows. The present study explores the influence of the rotation of nanoparticles on their luminescence anisotropy as a potential novel measurement method. This relation opens a new field of flow diagnostics, based on the measurement of polarized intensity components. Potentially, the method allows for the direct measurement of the vorticity. For this, the canonical flow in this study is a turbulent round jet at ${{Re}} = {12\,000}$ and 14 400. It is confirmed that the flow regime has an influence on the luminescence anisotropy. Using a model of such deterministic rotations according to another work by the authors (Schmidt & Rösgen, Phys. Rev. Res., vol. 5, no. 3, 2023, 033006), the magnitude of the vorticity components is computed, since the presented set-up is limited to sensing the magnitude of these quantities. The computed components indicate the self-similarity of the vorticity magnitude. A large-eddy simulation is conducted for comparison with the experiments, demonstrating good agreement.- 19 February 2024
- Published by Cambridge University Press (CUP) in Flow
- Vol. 4, E2
Open AccessLuminescence anisotropy and vorticity magnitude of a free turbulent jet
In the field of experimental fluid dynamics, the direct measurement of vorticity remains a challenge, even though it plays a crucial role in understanding turbulent flows. The present study explores the influence of the rotation of nanoparticles on their luminescence anisotropy as a potential novel measurement method. This relation opens a new field of flow diagnostics, based on the measurement of polarized intensity components. Potentially, the method allows for the direct measurement of the vorticity. For this, the canonical flow in this study is a turbulent round jet at ${{Re}} = {12\,000}$ and 14 400. It is confirmed that the flow regime has an influence on the luminescence anisotropy. Using a model of such deterministic rotations according to another work by the authors (Schmidt & Rösgen, Phys. Rev. Res., vol. 5, no. 3, 2023, 033006), the magnitude of the vorticity components is computed, since the presented set-up is limited to sensing the magnitude of these quantities. The computed components indicate the self-similarity of the vorticity magnitude. A large-eddy simulation is conducted for comparison with the experiments, demonstrating good agreement. - Open Access
- 11 December 2023
- Published by Springer Nature in Experiments in Fluids
- Vol. 65 (1) , 1-15
Simultaneous PIV–LIF measurements using RuPhen and a color camera
Particle image velocimetry (PIV) and laser-induced fluorescence (LIF) are currently state-of-the-art, non-intrusive measurement techniques that help in our understanding of heat and momentum transfer in thermal fluid flow applications. We present a simplified, integrated PIV–LIF system which simultaneously measures velocity and temperature fields in aqueous flows by means of a novel fluorescent dye (RuPhen), a low-cost continuous-wave diode laser, and a single color camera. We demonstrate that RuPhen is well-suited for this approach due to its peak absorption at 450 nm, peak emission at 605 nm, and a strong temperature-dependent emission with a sensitivity coefficient of $$4\%/^{\circ }\hbox {C}$$ . The large Stokes shift between excitation (which also includes Mie scattering of the flow tracers) and the emission facilitates the handling of the signal components in the RGB channels of the camera. To correct the recordings for laser power fluctuations, we propose a novel method that jointly employs two photoluminescence signals. We provide a spectral characterization of the dye at different temperatures and discuss the choice of each component for our measurement system. We demonstrate the potential of our approach by two experimental test cases that focus on thermally driven and turbulent flow regimes.- 11 December 2023
- Published by Springer Nature in Experiments in Fluids
- Vol. 65 (1) , 1-15
Open AccessSimultaneous PIV–LIF measurements using RuPhen and a color camera
Particle image velocimetry (PIV) and laser-induced fluorescence (LIF) are currently state-of-the-art, non-intrusive measurement techniques that help in our understanding of heat and momentum transfer in thermal fluid flow applications. We present a simplified, integrated PIV–LIF system which simultaneously measures velocity and temperature fields in aqueous flows by means of a novel fluorescent dye (RuPhen), a low-cost continuous-wave diode laser, and a single color camera. We demonstrate that RuPhen is well-suited for this approach due to its peak absorption at 450 nm, peak emission at 605 nm, and a strong temperature-dependent emission with a sensitivity coefficient of $$4\%/^{\circ }\hbox {C}$$ . The large Stokes shift between excitation (which also includes Mie scattering of the flow tracers) and the emission facilitates the handling of the signal components in the RGB channels of the camera. To correct the recordings for laser power fluctuations, we propose a novel method that jointly employs two photoluminescence signals. We provide a spectral characterization of the dye at different temperatures and discuss the choice of each component for our measurement system. We demonstrate the potential of our approach by two experimental test cases that focus on thermally driven and turbulent flow regimes. - Open Access
- 21 July 2023
- Published by Springer Nature in Experiments in Fluids
- Vol. 64 (8) , 1-9
TrackAER: real-time event-based quantitative flow visualization
We present a novel event-based quantitative flow visualization system, TrackAER, capable of continuously reconstructing, rendering and recording particle tracks in large test volumes without limitations on the measurement duration. Multiple event-based cameras are synchronized and calibrated to produce independent and asynchronous, yet temporally co-registered data streams of flow tracer positions. Subsequently, these data streams are merged into time-resolved three-dimensional particle tracks using photogrammetric techniques. Due to the operating principle of event cameras, the flow scenery is reduced to moving objects only, which effectively compresses the data stream at the camera source. In combination with an efficient data processing pipeline, the measurement system operates in real-time, reconstructing and rendering the investigated flow field without noticeable time lag. The data processing approach follows a “per-event” paradigm and enables the immediate observation and analysis of both, transient and long duration flow features. Specific issues resulting from event-based, frame-free processing are discussed as well as the advantages and limitations of event cameras. Exemplary results are provided to demonstrate the utility of the TrackAER system where Lagrangian particle track information is displayed in a virtual scene together with extracted quantitative information such as local flow velocities.
- 21 July 2023
- Published by Springer Nature in Experiments in Fluids
- Vol. 64 (8) , 1-9
Open AccessTrackAER: real-time event-based quantitative flow visualization
We present a novel event-based quantitative flow visualization system, TrackAER, capable of continuously reconstructing, rendering and recording particle tracks in large test volumes without limitations on the measurement duration. Multiple event-based cameras are synchronized and calibrated to produce independent and asynchronous, yet temporally co-registered data streams of flow tracer positions. Subsequently, these data streams are merged into time-resolved three-dimensional particle tracks using photogrammetric techniques. Due to the operating principle of event cameras, the flow scenery is reduced to moving objects only, which effectively compresses the data stream at the camera source. In combination with an efficient data processing pipeline, the measurement system operates in real-time, reconstructing and rendering the investigated flow field without noticeable time lag. The data processing approach follows a “per-event” paradigm and enables the immediate observation and analysis of both, transient and long duration flow features. Specific issues resulting from event-based, frame-free processing are discussed as well as the advantages and limitations of event cameras. Exemplary results are provided to demonstrate the utility of the TrackAER system where Lagrangian particle track information is displayed in a virtual scene together with extracted quantitative information such as local flow velocities.
- Open Access
- 5 July 2023
- Published by American Physical Society (APS) in Physical Review Research
- Vol. 5 (3) , 033006
Influence of deterministic rotations on luminescence anisotropy
The oriented emission of photons from an excited population of molecules, causing luminescence anisotropy, is subject to stochastic processes in a fluid. In addition to the well-known lateral Brownian motion, the population is exposed to random rotational processes as well, leading to a depolarization of the luminescence signal. In contrast, the rotation induced by a flow field, the vorticity, is a deterministic process. This rotation can be expressed in terms of a modified emission distribution function of the population. A model is developed to link these deterministic rotations to polarized intensity signals in an experiment. Analytical results of instantaneous and time-integrated intensity signals are provided, incorporating rotational Brownian motion and luminescence lifetime. The number of intensity signals is restricted to an experimentally feasible case, limiting the measurable results to the magnitude of vorticity components. The sensitivity regarding Brownian rotational motion and luminescence lifetime is evaluated. Next to the simplified solutions for 1- and 2-component vorticity fields, a minimization problem is presented to numerically solve the full problem of a 3-component vorticity field influencing the emission behavior. The developed Python toolbox is validated with a synthetic signals of a round turbulent jet.
- 5 July 2023
- Published by American Physical Society (APS) in Physical Review Research
- Vol. 5 (3) , 033006
Open AccessInfluence of deterministic rotations on luminescence anisotropy
The oriented emission of photons from an excited population of molecules, causing luminescence anisotropy, is subject to stochastic processes in a fluid. In addition to the well-known lateral Brownian motion, the population is exposed to random rotational processes as well, leading to a depolarization of the luminescence signal. In contrast, the rotation induced by a flow field, the vorticity, is a deterministic process. This rotation can be expressed in terms of a modified emission distribution function of the population. A model is developed to link these deterministic rotations to polarized intensity signals in an experiment. Analytical results of instantaneous and time-integrated intensity signals are provided, incorporating rotational Brownian motion and luminescence lifetime. The number of intensity signals is restricted to an experimentally feasible case, limiting the measurable results to the magnitude of vorticity components. The sensitivity regarding Brownian rotational motion and luminescence lifetime is evaluated. Next to the simplified solutions for 1- and 2-component vorticity fields, a minimization problem is presented to numerically solve the full problem of a 3-component vorticity field influencing the emission behavior. The developed Python toolbox is validated with a synthetic signals of a round turbulent jet.
- Open Access
- 1 July 2023
- Published by AIP Publishing in Physics of Fluids
- Vol. 35 (7)
Suppression of large-scale azimuthal modulations in a von Kármán flow using random forcing
A stable flow structure in a fully turbulent von Kármán swirling flow with counter-rotating disks is examined experimentally using global characterization methods (particle image velocimetry and ultrasonic profiling) and local methods at the center (ultrasonic Lagrangian tracking). The flow exhibits an azimuthal modulation in the central transverse plane, rotating slowly in the azimuthal direction. Contrary to previous studies, the bifurcation does not emerge from jumps of the shear layer plane but is symmetric to the transverse plane. To illustrate the underlying flow topology, a low-order model consisting of three superimposed modes is presented. A pseudo-random reversal of the disk rotation suppresses this dominant flow structure, retaining only the prototypical mean inward-pumping mode. Variations of the forcing method are studied for Reynolds numbers between 25 000 and 100 000, characteristic reversal times between 13 and 3000 Lagrangian integral times, and two reversal patterns. In contrast to a regular disk reversal, the employed pseudo-random sequence does not introduce any spurious timescales. The simple, yet efficient method is shown to robustly suppress the low-frequency signature of the azimuthal modulations over all Reynolds numbers under investigation. Globally, this yields a strong improvement in axisymmetric homogeneity and local statistical stationarity at moderate timescales. Also in the center of the cell, the removal of the highly anisotropic large-scale modulations enhances the transverse isotropy and homogeneity of the fully resolved turbulent flow and otherwise leaves the small-scale turbulent features largely unaffected. A description of the flow with the low-order model consequently reduces to only a single mode.
- 1 July 2023
- Published by AIP Publishing in Physics of Fluids
- Vol. 35 (7)
Open AccessSuppression of large-scale azimuthal modulations in a von Kármán flow using random forcing
A stable flow structure in a fully turbulent von Kármán swirling flow with counter-rotating disks is examined experimentally using global characterization methods (particle image velocimetry and ultrasonic profiling) and local methods at the center (ultrasonic Lagrangian tracking). The flow exhibits an azimuthal modulation in the central transverse plane, rotating slowly in the azimuthal direction. Contrary to previous studies, the bifurcation does not emerge from jumps of the shear layer plane but is symmetric to the transverse plane. To illustrate the underlying flow topology, a low-order model consisting of three superimposed modes is presented. A pseudo-random reversal of the disk rotation suppresses this dominant flow structure, retaining only the prototypical mean inward-pumping mode. Variations of the forcing method are studied for Reynolds numbers between 25 000 and 100 000, characteristic reversal times between 13 and 3000 Lagrangian integral times, and two reversal patterns. In contrast to a regular disk reversal, the employed pseudo-random sequence does not introduce any spurious timescales. The simple, yet efficient method is shown to robustly suppress the low-frequency signature of the azimuthal modulations over all Reynolds numbers under investigation. Globally, this yields a strong improvement in axisymmetric homogeneity and local statistical stationarity at moderate timescales. Also in the center of the cell, the removal of the highly anisotropic large-scale modulations enhances the transverse isotropy and homogeneity of the fully resolved turbulent flow and otherwise leaves the small-scale turbulent features largely unaffected. A description of the flow with the low-order model consequently reduces to only a single mode.
- Open Access
- 21 March 2023
- Published by Springer Nature in Experiments in Fluids
- Vol. 64 (4) , 1-16
A lightweight convolutional neural network to reconstruct deformation in BOS recordings
We introduce a Convolutional Neural Network (CNN) that is specifically designed and trained to post-process recordings obtained by Background Oriented Schlieren (BOS), a popular technique to visualize compressible and convective flows. To reconstruct BOS image deformation, we devised a lightweight network () that has comparatively fewer parameters to train than the CNNs that have been previously proposed for optical flow. To train , we introduce a novel strategy based on the generation of synthetic images from random-irrotational deformation fields, which are intended to mimic those provided by real BOS recordings. This allows us to generate a large number of training examples at minimal computational cost. To assess the accuracy of the reconstructed displacements, we consider test cases consisting of synthetic images with sinusoidal displacement as well as images obtained in the experimental studies of a hot plume in air and a flow past and inside a heated hollow hemisphere. By comparing the reconstructed deformation fields using the or conventional post-processing techniques used in Direct Image Correlation () or conventional image cross-correlation, we show that is more accurate and robust in the synthetic test case. When applied to experimental BOS recordings, all methods provide similar and consistent deformation fields. As is capable of achieving a comparable or better accuracy at a fraction of the computational costs, it represents a valuable alternative to conventional post-processing techniques for BOS experiments.
- 21 March 2023
- Published by Springer Nature in Experiments in Fluids
- Vol. 64 (4) , 1-16
Open AccessA lightweight convolutional neural network to reconstruct deformation in BOS recordings
We introduce a Convolutional Neural Network (CNN) that is specifically designed and trained to post-process recordings obtained by Background Oriented Schlieren (BOS), a popular technique to visualize compressible and convective flows. To reconstruct BOS image deformation, we devised a lightweight network () that has comparatively fewer parameters to train than the CNNs that have been previously proposed for optical flow. To train , we introduce a novel strategy based on the generation of synthetic images from random-irrotational deformation fields, which are intended to mimic those provided by real BOS recordings. This allows us to generate a large number of training examples at minimal computational cost. To assess the accuracy of the reconstructed displacements, we consider test cases consisting of synthetic images with sinusoidal displacement as well as images obtained in the experimental studies of a hot plume in air and a flow past and inside a heated hollow hemisphere. By comparing the reconstructed deformation fields using the or conventional post-processing techniques used in Direct Image Correlation () or conventional image cross-correlation, we show that is more accurate and robust in the synthetic test case. When applied to experimental BOS recordings, all methods provide similar and consistent deformation fields. As is capable of achieving a comparable or better accuracy at a fraction of the computational costs, it represents a valuable alternative to conventional post-processing techniques for BOS experiments.
- Open Access
- 4 January 2023
- Published by Springer Nature in Research Square
A lightweight convolutional neural network to reconstruct deformation in BOS recordings
We introduce a Convolutional Neural Network (CNN) that is specificallydesigned and trained to post-process recordings obtained by BackgroundOriented Schlieren (BOS), a popular technique to visualize compressibleand convective flows. To reconstruct BOS image deformation, we deviseda lightweight network (LIMA) that has comparatively fewer parameters totrain than the CNNs that have been previously proposed for optical flow.To train LIMA, we introduce a novel strategy based on the generation of synthetic images from random-irrotational deformation fields, which areintended to mimic those provided by real BOS recordings. This allows us to generate a large number of training examples at minimal compu-tational cost. To assess the accuracy of the reconstructed displacements,we consider test cases consisting of synthetic images with sinusoidal dis-placement as well as images obtained in the experimental studies of a hotplume in air and a flow past and inside a heated hollow hemisphere. By comparing the reconstructed deformation fields using the LIMA or conven-tional post-processing techniques used in Direct Image Correlation (dic)or conventional image cross-correlation, we show that LIMA is more accu-rate and robust in the synthetic test case. When applied to experimentalBOS recordings, all methods provide similar and consistent deformation fields. As LIMA is capable of achieving a comparable or better accuracyat a fraction of the computational costs, it represents a valuable alternative to conventional post-processing techniques for BOS experiments.
- 4 January 2023
- Published by Springer Nature in Research Square
Open AccessA lightweight convolutional neural network to reconstruct deformation in BOS recordings
We introduce a Convolutional Neural Network (CNN) that is specificallydesigned and trained to post-process recordings obtained by BackgroundOriented Schlieren (BOS), a popular technique to visualize compressibleand convective flows. To reconstruct BOS image deformation, we deviseda lightweight network (LIMA) that has comparatively fewer parameters totrain than the CNNs that have been previously proposed for optical flow.To train LIMA, we introduce a novel strategy based on the generation of synthetic images from random-irrotational deformation fields, which areintended to mimic those provided by real BOS recordings. This allows us to generate a large number of training examples at minimal compu-tational cost. To assess the accuracy of the reconstructed displacements,we consider test cases consisting of synthetic images with sinusoidal dis-placement as well as images obtained in the experimental studies of a hotplume in air and a flow past and inside a heated hollow hemisphere. By comparing the reconstructed deformation fields using the LIMA or conven-tional post-processing techniques used in Direct Image Correlation (dic)or conventional image cross-correlation, we show that LIMA is more accu-rate and robust in the synthetic test case. When applied to experimentalBOS recordings, all methods provide similar and consistent deformation fields. As LIMA is capable of achieving a comparable or better accuracyat a fraction of the computational costs, it represents a valuable alternative to conventional post-processing techniques for BOS experiments.
- Open Access
- 24 August 2022
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 33 (11) , 114005
Deep learning based instance segmentation of particle streaks and tufts
3D particle streak velocimetry (3D-PSV) and surface flow visualization using tufts both require the detection of curve segments, particle streaks or tufts, in images. We propose the use of deep learning based instance segmentation neural networks Mask region-based convolutional neural network (R-CNN) and Cascade Mask R-CNN, trained on fully synthetic data, to accurately identify, segment, and classify streaks and tufts. For 3D-PSV, we use the segmented masks and detected streak endpoints to volumetrically reconstruct flows even when the imaged streaks partly overlap or intersect. In addition, we use Mask R-CNN to segment images of tufts and classify the detected tufts according to their range of motion, thus automating the detection of regions of separated flow while at the same time providing accurate segmentation masks. Finally, we show a successful synthetic-to-real transfer by training only on synthetic data and successfully evaluating real data. The synthetic data generation is particularly suitable for the two presented applications, as the experimental images consist of simple geometric curves or a superposition of curves. Therefore, the proposed networks provide a general framework for instance detection, keypoint detection and classification that can be fine-tuned to the specific experimental application and imaging parameters using synthetic data.
- 24 August 2022
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 33 (11) , 114005
Open AccessDeep learning based instance segmentation of particle streaks and tufts
3D particle streak velocimetry (3D-PSV) and surface flow visualization using tufts both require the detection of curve segments, particle streaks or tufts, in images. We propose the use of deep learning based instance segmentation neural networks Mask region-based convolutional neural network (R-CNN) and Cascade Mask R-CNN, trained on fully synthetic data, to accurately identify, segment, and classify streaks and tufts. For 3D-PSV, we use the segmented masks and detected streak endpoints to volumetrically reconstruct flows even when the imaged streaks partly overlap or intersect. In addition, we use Mask R-CNN to segment images of tufts and classify the detected tufts according to their range of motion, thus automating the detection of regions of separated flow while at the same time providing accurate segmentation masks. Finally, we show a successful synthetic-to-real transfer by training only on synthetic data and successfully evaluating real data. The synthetic data generation is particularly suitable for the two presented applications, as the experimental images consist of simple geometric curves or a superposition of curves. Therefore, the proposed networks provide a general framework for instance detection, keypoint detection and classification that can be fine-tuned to the specific experimental application and imaging parameters using synthetic data.