Robin Leister
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Robin Leister has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
- Jochen Kriegseis
Karlsruhe Institute of Technology (KIT)
12 shared publications - Alexander Stroh
Institute of Fluid Mechanics (ISTM), Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany
4 shared publications - Maximilian Dreisbach
Institute of Fluids Mechanics (ISTM), Karlsruhe Institute of Technology (KIT), Kaiserstraße 10, 76131 Karlsruhe, Germany
4 shared publications - Bettina Frohnapfel
Karlsruhe Institute of Technology
3 shared publications - Saskia Pasch
Institute of Fluids Mechanics (ISTM), Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany
3 shared publications - Christian Sax
Institute of Fluids Mechanics (ISTM),Karlsruhe Institute of Technology (KIT), Kaiserstraße 10, 76131 Karlsruhe, Germany
2 shared publications
Latest Publications
- Open Access
- 30 August 2025
- Published by Springer Nature in Experiments in Fluids
- Vol. 66 (9) , 1-18
Defocusing PTV flow measurements above a DBD plasma actuator array for oscillatory forcing
Lagrangian defocusing particle tracking velocimetry (defocusing PTV, DPTV) measurements are performed in a thin volume above a plasma actuator array that is applied to mimic the effect of wall oscillations by inducing alternating, wall-parallel forcing in opposite directions into the air above the actuator surface for flow control purposes. The aim of the experiments is to capture the plasma-induced flow topology in otherwise quiescent air throughout the oscillation cycle within the measurement volume of 14 mm $$\times$$ 1 mm $$\times$$ 14 mm, immediately adjacent to the wall-mounted actuator. For this purpose, particle image velocimetry equipment for time-resolved measurements with one camera is used in a DPTV setup, where the out-of-plane particle coordinate is obtained through the diameter of a defocused particle image. Three-dimensional, three-component velocity and acceleration data is extracted by introducing a continuous particle tracking approach and an extended ex situ calibration procedure based on the detection of solid particles directly applied to a wall boundary, for which no prior knowledge of the flow topology or velocity data in the direct vicinity of the wall is required. A novel method for estimating measurement uncertainty in this context is introduced, and the influencing factors are discussed from an application perspective. Through the analysis of Lagrangian particle tracks, both individual flow events and statistical effects within the oscillation cycle can be evaluated. The extraction of phase-resolved flow fields with adaptable spatial resolution shows the forcing effect to be regular across different discharge zones on the plasma actuator array, indicating well-balanced voltage settings and precise manufacturing. Furthermore, the relation between the forcing-induced velocity and acceleration fields is quantitatively assessed, revealing the spatio-temporal transmission characteristics of the applied forcing. In summary, the obtained results demonstrate the applicability of DPTV measurement technique for the flow characterization above a plasma actuator array using the presented modifications.- 30 August 2025
- Published by Springer Nature in Experiments in Fluids
- Vol. 66 (9) , 1-18
Open AccessDefocusing PTV flow measurements above a DBD plasma actuator array for oscillatory forcing
Lagrangian defocusing particle tracking velocimetry (defocusing PTV, DPTV) measurements are performed in a thin volume above a plasma actuator array that is applied to mimic the effect of wall oscillations by inducing alternating, wall-parallel forcing in opposite directions into the air above the actuator surface for flow control purposes. The aim of the experiments is to capture the plasma-induced flow topology in otherwise quiescent air throughout the oscillation cycle within the measurement volume of 14 mm $$\times$$ 1 mm $$\times$$ 14 mm, immediately adjacent to the wall-mounted actuator. For this purpose, particle image velocimetry equipment for time-resolved measurements with one camera is used in a DPTV setup, where the out-of-plane particle coordinate is obtained through the diameter of a defocused particle image. Three-dimensional, three-component velocity and acceleration data is extracted by introducing a continuous particle tracking approach and an extended ex situ calibration procedure based on the detection of solid particles directly applied to a wall boundary, for which no prior knowledge of the flow topology or velocity data in the direct vicinity of the wall is required. A novel method for estimating measurement uncertainty in this context is introduced, and the influencing factors are discussed from an application perspective. Through the analysis of Lagrangian particle tracks, both individual flow events and statistical effects within the oscillation cycle can be evaluated. The extraction of phase-resolved flow fields with adaptable spatial resolution shows the forcing effect to be regular across different discharge zones on the plasma actuator array, indicating well-balanced voltage settings and precise manufacturing. Furthermore, the relation between the forcing-induced velocity and acceleration fields is quantitatively assessed, revealing the spatio-temporal transmission characteristics of the applied forcing. In summary, the obtained results demonstrate the applicability of DPTV measurement technique for the flow characterization above a plasma actuator array using the presented modifications. - Open Access
- 28 June 2025
- Published by Springer Nature in Experiments in Fluids
- Vol. 66 (7) , 1-12
Escaping the lab: advances in large-scale particle tracking using natural light and uncrewed aerial vehicles
Here we explore the potential of the glare-point particle tracking (GPPT) technique for realistic field-scale measurements. We make use of a commercially available, drone-based camera to extract three-dimensional information under natural light conditions. Air-filled soap bubbles on the order of centimeters are used as seeding tracers. In the current tests, the suitability of the portable setup has been demonstrated for volumes up to $$163\,$$ m $$^3$$ . The frame-to-frame camera movement, caused by slight adjustments of the drone flying outdoors, could be quantified and corrected via an image-based approach. Both limits introduced through the larger tracer size and uncertainty caused by the glare-point approach are discussed accordingly. Furthermore, based on the fixed magnification of the drone camera, the limit in drone operation can be determined when the two most dominant, bubble-based glare points collapse onto one, which was observed well above a height of $$10\,$$ m. Finally, a turbulent free jet, exiting from a square-shaped nozzle with edge length $$0.3\,$$ m, served as an exemplary test case. Tracks emerging at the jet exit all the way to $$10\,$$ m downstream of the nozzle were successfully reconstructed in three dimensions. Lagrangian properties, such as flow acceleration, as well as pathline curvature at the integral length scale, were resolved in the large turbulent jet. This sparse yet accurate Lagrangian data over a large physical volume demonstrates insights into turbulent mixing processes at the integral length well beyond current Eulerian-based descriptions.- 28 June 2025
- Published by Springer Nature in Experiments in Fluids
- Vol. 66 (7) , 1-12
Open AccessEscaping the lab: advances in large-scale particle tracking using natural light and uncrewed aerial vehicles
Here we explore the potential of the glare-point particle tracking (GPPT) technique for realistic field-scale measurements. We make use of a commercially available, drone-based camera to extract three-dimensional information under natural light conditions. Air-filled soap bubbles on the order of centimeters are used as seeding tracers. In the current tests, the suitability of the portable setup has been demonstrated for volumes up to $$163\,$$ m $$^3$$ . The frame-to-frame camera movement, caused by slight adjustments of the drone flying outdoors, could be quantified and corrected via an image-based approach. Both limits introduced through the larger tracer size and uncertainty caused by the glare-point approach are discussed accordingly. Furthermore, based on the fixed magnification of the drone camera, the limit in drone operation can be determined when the two most dominant, bubble-based glare points collapse onto one, which was observed well above a height of $$10\,$$ m. Finally, a turbulent free jet, exiting from a square-shaped nozzle with edge length $$0.3\,$$ m, served as an exemplary test case. Tracks emerging at the jet exit all the way to $$10\,$$ m downstream of the nozzle were successfully reconstructed in three dimensions. Lagrangian properties, such as flow acceleration, as well as pathline curvature at the integral length scale, were resolved in the large turbulent jet. This sparse yet accurate Lagrangian data over a large physical volume demonstrates insights into turbulent mixing processes at the integral length well beyond current Eulerian-based descriptions. - Open Access
- 6 January 2025
- Published by Springer Nature in Cardiovascular Engineering and Technology
- Vol. 16 (2) , 155-170
Investigating the Shortcomings of the Flow Convergence Method for Quantification of Mitral Regurgitation in a Pulsatile In-Vitro Environment and with Computational Fluid Dynamics
The flow convergence method includes calculation of the proximal isovelocity surface area (PISA) and is widely used to classify mitral regurgitation (MR) with echocardiography. It constitutes a primary decision factor for determination of treatment and should therefore be a robust quantification method. However, it is known for its tendency to underestimate MR and its dependence on user expertise. The present work systematically compares different pulsatile flow profiles arising from different regurgitation orifices using transesophageal echocardiographic (TEE) probe and particle image velocimetry (PIV) as a reference in an in-vitro environment. It is found that the inter-observer variability using echocardiography is small compared to the systematic underestimation of the regurgitation volume for large orifice areas (up to 52%) where a violation of the flow convergence method assumptions occurs. From a flow perspective, a starting vortex was found as a dominant flow pattern in the regurgant jet for all orifice shapes and sizes. A series of simplified computational fluid dynamics (CFD) simulations indicate that selecting a suboptimal aliasing velocity during echocardiography measurements might be a primary source of potential underestimation in MR characterization via the PISA-based method, reaching up to 40%. In this study, it has been noted in clinical observations that physicians often select an aliasing velocity higher than necessary for optimal estimation in diagnostic procedures.
- 6 January 2025
- Published by Springer Nature in Cardiovascular Engineering and Technology
- Vol. 16 (2) , 155-170
Open AccessInvestigating the Shortcomings of the Flow Convergence Method for Quantification of Mitral Regurgitation in a Pulsatile In-Vitro Environment and with Computational Fluid Dynamics
The flow convergence method includes calculation of the proximal isovelocity surface area (PISA) and is widely used to classify mitral regurgitation (MR) with echocardiography. It constitutes a primary decision factor for determination of treatment and should therefore be a robust quantification method. However, it is known for its tendency to underestimate MR and its dependence on user expertise. The present work systematically compares different pulsatile flow profiles arising from different regurgitation orifices using transesophageal echocardiographic (TEE) probe and particle image velocimetry (PIV) as a reference in an in-vitro environment. It is found that the inter-observer variability using echocardiography is small compared to the systematic underestimation of the regurgitation volume for large orifice areas (up to 52%) where a violation of the flow convergence method assumptions occurs. From a flow perspective, a starting vortex was found as a dominant flow pattern in the regurgant jet for all orifice shapes and sizes. A series of simplified computational fluid dynamics (CFD) simulations indicate that selecting a suboptimal aliasing velocity during echocardiography measurements might be a primary source of potential underestimation in MR characterization via the PISA-based method, reaching up to 40%. In this study, it has been noted in clinical observations that physicians often select an aliasing velocity higher than necessary for optimal estimation in diagnostic procedures.
- Open Access
- 29 February 2024
- Published by Springer Nature in Flow, Turbulence and Combustion
- Vol. 113 (1) , 215-216
Correction to: Measurements in a Turbulent Channel Flow by Means of an LDV Profile Sensor
- 29 February 2024
- Published by Springer Nature in Flow, Turbulence and Combustion
- Vol. 113 (1) , 215-216
Open AccessCorrection to: Measurements in a Turbulent Channel Flow by Means of an LDV Profile Sensor
- 24 October 2023
- Published by Cassyni
Defocusing PTV applied to an open wet clutch
The optical measurement technique Defocusing Particle Tracking Velocimetry (Defocusing PTV) is applied to the sub-millimeter gap of an open wet clutch to gain deeper insights into the unknown flow, which causes a significant loss in nowadays automobiles. The present work improves the fundamental understanding of the flow and its contribution to the generated drag torque and the physical process of aeration. A set of governing analytical equations is revealed from in-depth theoretical considerations, which describe the general cause-effect relations of the flow. To gain deeper insights into the unknown intra-groove phenomena Defocusing PTV is successfully applied to (locally) extract precise vortex information and fine resolved wall shear stress values. Metrological insights are generated with the introduction of a new detection strategy and the proven flexibility of Defocusing PTV, which makes comprehensive magnification and location-accuracy studies possible. The work is completed with a flow analysis along the entire radial region of interest, and the consideration of a more complex groove geometry.
- 24 October 2023
- Published by Cassyni
Defocusing PTV applied to an open wet clutch
The optical measurement technique Defocusing Particle Tracking Velocimetry (Defocusing PTV) is applied to the sub-millimeter gap of an open wet clutch to gain deeper insights into the unknown flow, which causes a significant loss in nowadays automobiles. The present work improves the fundamental understanding of the flow and its contribution to the generated drag torque and the physical process of aeration. A set of governing analytical equations is revealed from in-depth theoretical considerations, which describe the general cause-effect relations of the flow. To gain deeper insights into the unknown intra-groove phenomena Defocusing PTV is successfully applied to (locally) extract precise vortex information and fine resolved wall shear stress values. Metrological insights are generated with the introduction of a new detection strategy and the proven flexibility of Defocusing PTV, which makes comprehensive magnification and location-accuracy studies possible. The work is completed with a flow analysis along the entire radial region of interest, and the consideration of a more complex groove geometry.
- Open Access
- 11 September 2023
- Published by Springer Nature in Flow, Turbulence and Combustion
- Vol. 113 (1) , 195-213
Measurements in a Turbulent Channel Flow by Means of an LDV Profile Sensor
Spatially and temporally resolved velocity measurements in wall-bounded turbulent flows remain a challenge. Contrary to classical laser Doppler velocimetry (LDV) measurements, the laser Doppler velocity profile sensor (LDV-PS) allows the combined measurement of tracer particle position and velocity, which makes it a promising tool. To assess its feasibility a commercial LDV-PS is employed in a turbulent channel flow at $$Re_\tau =350$$ . Additionally, the measurement and signal-processing accuracies of velocity and location are evaluated for various tracer-object sizes and velocities. On this basis, the turbulent channel flow measurements are evaluated and compared to reference data from direct numerical simulations. Thus, potentials of the LDV-PS are investigated for different regions of the flow and various data processing routines as well as the experimental practice are discussed from an application perspective.- 11 September 2023
- Published by Springer Nature in Flow, Turbulence and Combustion
- Vol. 113 (1) , 195-213
Open AccessMeasurements in a Turbulent Channel Flow by Means of an LDV Profile Sensor
Spatially and temporally resolved velocity measurements in wall-bounded turbulent flows remain a challenge. Contrary to classical laser Doppler velocimetry (LDV) measurements, the laser Doppler velocity profile sensor (LDV-PS) allows the combined measurement of tracer particle position and velocity, which makes it a promising tool. To assess its feasibility a commercial LDV-PS is employed in a turbulent channel flow at $$Re_\tau =350$$ . Additionally, the measurement and signal-processing accuracies of velocity and location are evaluated for various tracer-object sizes and velocities. On this basis, the turbulent channel flow measurements are evaluated and compared to reference data from direct numerical simulations. Thus, potentials of the LDV-PS are investigated for different regions of the flow and various data processing routines as well as the experimental practice are discussed from an application perspective. - Open Access
- 5 September 2023
- Published by Springer Nature in Forschung im Ingenieurwesen
- Vol. 87 (4) , 1297-1306
Fluid-mechanical evaluation of different clutch geometries based on experimental and numerical investigations
Wet clutches in their open state add losses caused by drag torque to the drive train, making the optimization of the disk design and drag torque reduction a core development aspect. The present work focuses on the influence of the chosen disk-groove geometry on the resulting flow topology in open wet clutches. Therefore, the flow topology of six different disk designs is investigated experimentally and numerically. Other influences of the operating conditions such as volume flow or other design elements such as wave springs are not considered. New parameters for the flow topology are derived, for a better description of the influence of the flow topology on the drag torque. Based on these insights strategies for further understanding of the complex flow topology on open wet clutches are derived and optimization approaches proposed.
- 5 September 2023
- Published by Springer Nature in Forschung im Ingenieurwesen
- Vol. 87 (4) , 1297-1306
Open AccessFluid-mechanical evaluation of different clutch geometries based on experimental and numerical investigations
Wet clutches in their open state add losses caused by drag torque to the drive train, making the optimization of the disk design and drag torque reduction a core development aspect. The present work focuses on the influence of the chosen disk-groove geometry on the resulting flow topology in open wet clutches. Therefore, the flow topology of six different disk designs is investigated experimentally and numerically. Other influences of the operating conditions such as volume flow or other design elements such as wave springs are not considered. New parameters for the flow topology are derived, for a better description of the influence of the flow topology on the drag torque. Based on these insights strategies for further understanding of the complex flow topology on open wet clutches are derived and optimization approaches proposed.
- Open Access
- 29 June 2023
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 34 (9) , 095909
Deep learning and hybrid approach for particle detection in defocusing particle tracking velocimetry
The present work aims at the improvement of particle detection in defocusing particle tracking velocimetry (DPTV) by means of a novel hybrid approach. Two deep learning approaches, namely faster R-CNN and RetinaNet are compared to the performance of two benchmark conventional image processing algorithms for DPTV. For the development of a hybrid approach with improved performance, the different detection approaches are evaluated on synthetic and images from an actual DPTV experiment. First, the performance under the influence of noise, overlaps, seeding density and optical aberrations is discussed and consequently advantages of neural networks over conventional image processing algorithms for image processing in DPTV are derived. Furthermore, current limitations of the application of neural networks for DPTV are pointed out and their origin is elaborated. It shows that neural networks have a better detection capability but suffer from low positional accuracy when locating particles. Finally, a novelHybrid Approachis proposed, which uses a neural network for particle detection and passes the prediction onto a conventional refinement algorithm for better position accuracy. A third step is implemented to additionally eliminate false predictions by the network based on a subsequent rejection criterion. The novel approach improves the powerful detection performance of neural networks while maintaining the high position accuracy of conventional algorithms, combining the advantages of both approaches.
- 29 June 2023
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 34 (9) , 095909
Open AccessDeep learning and hybrid approach for particle detection in defocusing particle tracking velocimetry
The present work aims at the improvement of particle detection in defocusing particle tracking velocimetry (DPTV) by means of a novel hybrid approach. Two deep learning approaches, namely faster R-CNN and RetinaNet are compared to the performance of two benchmark conventional image processing algorithms for DPTV. For the development of a hybrid approach with improved performance, the different detection approaches are evaluated on synthetic and images from an actual DPTV experiment. First, the performance under the influence of noise, overlaps, seeding density and optical aberrations is discussed and consequently advantages of neural networks over conventional image processing algorithms for image processing in DPTV are derived. Furthermore, current limitations of the application of neural networks for DPTV are pointed out and their origin is elaborated. It shows that neural networks have a better detection capability but suffer from low positional accuracy when locating particles. Finally, a novelHybrid Approachis proposed, which uses a neural network for particle detection and passes the prediction onto a conventional refinement algorithm for better position accuracy. A third step is implemented to additionally eliminate false predictions by the network based on a subsequent rejection criterion. The novel approach improves the powerful detection performance of neural networks while maintaining the high position accuracy of conventional algorithms, combining the advantages of both approaches.
- Open Access
- 1 May 2023
- Published by Springer Nature in Experiments in Fluids
- Vol. 64 (5) , 1-9
Defocusing PTV applied to an open wet clutch: from macro to micro
Defocusing particle tracking velocimetry (defocusing PTV) is applied in various levels of magnification to investigate the flow in an open wet clutch with radial grooving. The magnifications range from $$M=0.8$$ up to $$M=15$$ , whereas the corresponding defocusing sensitivities $$s^{Z}$$ , denoting the rate of the particle image diameter change along the optical axis, range from 67.1 to 9.0 µm/pixel. For flows with sub-millimeter structures, as it is the case of this clutch flow application, a high spatial resolution is necessary to fully understand the flow phenomena in these confined geometries. It is demonstrated in this study that all optical configurations can resolve the miniature flow structures that are present in an open wet clutch. Moreover, the $$M=0.8$$ set-up, featuring the largest field of view, captures the entire region between the inner and the outer radius of the clutch, providing a comprehensive overview of the three-dimensional clutch flow field, ranging from the smallest structures to the largest. It is found that the $$M=0.8$$ set-up also yields the lowest uncertainties in the determination of the particle locations in x, y direction as well as in z direction along the optical axis. The particle image diameter uncertainty is $$2\sigma _{d_{\textrm{i}}}=0.24\,$$ pixel, which translates (by multiplying the defocusing sensitivity of $$s^{Z}=67.1\,$$ µm/pixel) into a particle z location determination uncertainty of $$2\sigma _{z}=16.1\,$$ µm. These results give evidence to the fact that not necessarily the best defocusing sensitivity determines the accuracy of the particle z location estimation; to a great extent the signal-to-noise ratio and the width of the outer ring of the defocused particle image influence the accuracy. Graphical Abstract- 1 May 2023
- Published by Springer Nature in Experiments in Fluids
- Vol. 64 (5) , 1-9
Open AccessDefocusing PTV applied to an open wet clutch: from macro to micro
Defocusing particle tracking velocimetry (defocusing PTV) is applied in various levels of magnification to investigate the flow in an open wet clutch with radial grooving. The magnifications range from $$M=0.8$$ up to $$M=15$$ , whereas the corresponding defocusing sensitivities $$s^{Z}$$ , denoting the rate of the particle image diameter change along the optical axis, range from 67.1 to 9.0 µm/pixel. For flows with sub-millimeter structures, as it is the case of this clutch flow application, a high spatial resolution is necessary to fully understand the flow phenomena in these confined geometries. It is demonstrated in this study that all optical configurations can resolve the miniature flow structures that are present in an open wet clutch. Moreover, the $$M=0.8$$ set-up, featuring the largest field of view, captures the entire region between the inner and the outer radius of the clutch, providing a comprehensive overview of the three-dimensional clutch flow field, ranging from the smallest structures to the largest. It is found that the $$M=0.8$$ set-up also yields the lowest uncertainties in the determination of the particle locations in x, y direction as well as in z direction along the optical axis. The particle image diameter uncertainty is $$2\sigma _{d_{\textrm{i}}}=0.24\,$$ pixel, which translates (by multiplying the defocusing sensitivity of $$s^{Z}=67.1\,$$ µm/pixel) into a particle z location determination uncertainty of $$2\sigma _{z}=16.1\,$$ µm. These results give evidence to the fact that not necessarily the best defocusing sensitivity determines the accuracy of the particle z location estimation; to a great extent the signal-to-noise ratio and the width of the outer ring of the defocused particle image influence the accuracy. Graphical Abstract - Open Access
- 8 September 2022
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 33 (12) , 124001
Particle detection by means of neural networks and synthetic training data refinement in defocusing particle tracking velocimetry
The presented work addresses the problem of particle detection with neural networks (NNs) in defocusing particle tracking velocimetry. A novel approach based on synthetic training data refinement is introduced, with the scope of revising the well documented performance gap of synthetically trained NNs, applied to experimental recordings. In particular, synthetic particle image (PI) data is enriched with image features from the experimental recordings by means of deep learning through an unsupervised image-to-image translation. It is demonstrated that this refined synthetic training data enables the neural-network-based particle detection for a simultaneous increase in detection rate and reduction in the rate of false positives, beyond the capability of conventional detection algorithms. The potential for an increased accuracy in particle detection is revealed with NNs that utilise small scale image features, which further underlines the importance of representative training data. In addition, it is demonstrated that NNs are able to resolve overlapping PIs with a higher reliability and accuracy in comparison to conventional algorithms, suggesting the possibility of an increased seeding density in real experiments. A further finding is the robustness of NNs to inhomogeneous background illumination and aberration of the images, which opens up defocusing PTV for a wider range of possible applications. The successful application of synthetic training-data refinement advances the neural-network-based particle detection towards real world applicability and suggests the potential of a further performance gain from more suitable training data.
- 8 September 2022
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 33 (12) , 124001
Open AccessParticle detection by means of neural networks and synthetic training data refinement in defocusing particle tracking velocimetry
The presented work addresses the problem of particle detection with neural networks (NNs) in defocusing particle tracking velocimetry. A novel approach based on synthetic training data refinement is introduced, with the scope of revising the well documented performance gap of synthetically trained NNs, applied to experimental recordings. In particular, synthetic particle image (PI) data is enriched with image features from the experimental recordings by means of deep learning through an unsupervised image-to-image translation. It is demonstrated that this refined synthetic training data enables the neural-network-based particle detection for a simultaneous increase in detection rate and reduction in the rate of false positives, beyond the capability of conventional detection algorithms. The potential for an increased accuracy in particle detection is revealed with NNs that utilise small scale image features, which further underlines the importance of representative training data. In addition, it is demonstrated that NNs are able to resolve overlapping PIs with a higher reliability and accuracy in comparison to conventional algorithms, suggesting the possibility of an increased seeding density in real experiments. A further finding is the robustness of NNs to inhomogeneous background illumination and aberration of the images, which opens up defocusing PTV for a wider range of possible applications. The successful application of synthetic training-data refinement advances the neural-network-based particle detection towards real world applicability and suggests the potential of a further performance gain from more suitable training data.