SerialTrack: ScalE and rotation invariant augmented Lagrangian particle tracking

Yang, Jin; Yin, Yimu; Landauer, Alexander K.; Buyuktozturk, Selda; Zhang, Jing; Summey, Luke; McGhee, A.; Fu, Matthew; Dabiri, John O.; Franck, Christian

doi:10.1016/j.softx.2022.101204

Cited by 6 publications

(5 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hence, testing on real-world datasets becomes essential for validation. To ascertain our method's efficacy in real-world PTV experiments, in this section, we test it with two real-world datasets, DeformationFlow [32] and Aortic Valve Interstitial Cell (AVIC) [33]. The models, trained on the FluidFlow3D-NORM dataset, are utilized for two distinct real-world PTV tasks: one in the physical domain and the other in the biological domain.…”

Section: Real Ptv Experimentsmentioning

confidence: 99%

“…Therefore, we adopt the existing PTV algorithm as a framework and use our method as its initialisation. We employ SerialTrack [32] as the particle tracking framework for this experiment. The original version is denoted as SerialTrack, while its adaptation with our scene flow estimation framework is termed Ours+ST.…”

Section: Deformationflowmentioning

confidence: 99%

“…The DeformationFlow dataset [32] showcases real-world physical dynamics by recording the indentation created when a stainless steel sphere is placed on a soft polyacrylamide hydrogel due to gravitational forces. Fluorescent fluid particles within the gel were scanned both pre and post-indentation, yielding 3D volumetric images.…”

Section: Deformationflowmentioning

confidence: 99%

See 2 more Smart Citations

Test-time Self Supervision for Data Efficient and Cross-domain Robust Particle Tracking in Turbulent Flow

Zhang,

Gao,

Jiang

et al. 2023

Preprint

View full text Add to dashboard Cite

3D particle tracking velocimetry (PTV), which allows the determination of instantaneous velocity, is a key technique for analyzing 3D turbulent flow. While deep learning based methods have fostered advancements in the realm of 3D PTV, they remain heavily dependent on vast volumes of labeled data.This dependency is problematic considering the challenges of collecting particle sets in a large scale, especially when the data collection process is complex or the data itself is rare.Additionally, supervised learning excels primarily with in-domain data, and it may not exhibit the same efficacy with out-of-domain samples.To improve data efficiency and robustness, this paper introduces an innovative test-time self-supervised framework designed for the task of 3D PTV.Notably, we consider an extremely challenging setting of using only 1% data that previous methods use.However, drastically reducing the training samples to this extent can severely hurt performance. To alleviate this, we introduce the zero-divergence loss as an additional regularization mechanism, which draws inspiration from the inherent zero-divergence principle of incompressible fluid's velocity fields.To further enhance cross-domain robustness, we introduce a module named Dynamic Velocimetry Enhancer. During test-time, this module leverages a reconstruction loss to optimize the initial flow predicted by a trained network, enabling the estimated flow to better adjust to unseen samples.Through comprehensive experiments, we reveal that our novel test-time self-supervised framework notably outperforms fully-supervised counterparts with only 1% training samples used by previous methods.Additionally, our cross-domain robustness analysis confirms our framework can naturally generalize to new cases at the merit of test-time optimization. We finally present two evaluations on real-world data in the physical and biological domains, respectively, highlighting the valuable potential for applications in new, noisy real-world scenarios.

show abstract

Section: Real Ptv Experimentsmentioning

confidence: 99%

Section: Deformationflowmentioning

confidence: 99%

See 1 more Smart Citation

Test-time Self Supervision for Data Efficient and Cross-domain Robust Particle Tracking in Turbulent Flow

Zhang,

Gao,

Jiang

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Randomly sampled data poses a challenge to gradient estimation. Even though local least-squares fitting of displacement fields and interpolation from irregular data to a regular grid have been reported [ 45 , 56 ], performance of the methods under large gradients is unclear, and the related estimation errors have not been completely characterized.…”

Section: Introductionmentioning

confidence: 99%

“…In this manuscript, a full-field method for estimating the deformation gradient tensor is proposed, based on the non-gridded displacement data obtained from particle tracking in a manner similar to recent work [ 56 ]. At each particle location, the deformation gradient tensor is estimated from the displacement measurements of the given center particle and its k -nearest neighbors (see the inset of Fig.…”

Section: Introductionmentioning

confidence: 99%

Estimation of the Deformation Gradient Tensor by Particle Tracking Near a Free Boundary with Quantified Error

Benkley,

Li,

Kolinski

2023

Exp Mech

View full text Add to dashboard Cite

Background Obtaining accurate displacement measurements for large material deformation and/or rotation presents a distinct challenge to digital image correlation (DIC) due to cumulative and decorrelation errors, particularly near material boundaries. Objective We aim to accurately measure the deformation gradient tensor near boundary discontinuities in situations of large deformation and large deformation gradients. Methods To achieve this goal, the locations of randomly distributed particles are tracked using an open-source particle-tracking software, Trackpy. A least-squares estimate of the deformation gradient tensor field uses nearest-neighbor material vectors and a first-order Finite Difference (FD) approximation, circumventing common errors in other methods. The error caused by FD approximation and that incurred by measurement are derived and tested with exhaustive numerical simulations. Furthermore, a uniaxial tensile test and mode-I fracture experiment are conducted with particle-embedded hydrogels to validate the method. Results Numerical results corroborate a theoretical expression of measurement error. They show that the FD error increases while the measurement error decreases for a growing estimating radius. Moreover, measurement error is linearly correlated to displacement noise. A benchmark uniaxial tensile test validates the accuracy of the proposed estimator, and the near-crack-tip measurements in a tensile fracture experiment demonstrate the estimator’s capabilities near a free surface, when a material undergoes large deformation and rotation. The results of the displacement and strain data are benchmarked against kinematic data obtained using an open-source DIC software, Ncorr. Computation time for both methods is compared. Conclusions A deformation gradient tensor estimator is developed based on a particle tracking technique and a least squares routine. Theoretical error bounds on the estimator are verified by numerical simulations, and the method’s capability is confirmed by physical experiments in evaluating large deformation and rotation near a free boundary. The proposed estimator is expected to open a door towards future material tests and experimental mechanics studies, especially in large deformation and large rotation scenarios.

show abstract

High-Speed, Full-Field Deformation Measurements Near Inertial Microcavitation Bubbles Inside Viscoelastic Hydrogels

et al. 2022

View full text Add to dashboard Cite

SerialTrack: ScalE and rotation invariant augmented Lagrangian particle tracking

Cited by 6 publications

References 56 publications

Test-time Self Supervision for Data Efficient and Cross-domain Robust Particle Tracking in Turbulent Flow

Test-time Self Supervision for Data Efficient and Cross-domain Robust Particle Tracking in Turbulent Flow

Estimation of the Deformation Gradient Tensor by Particle Tracking Near a Free Boundary with Quantified Error

High-Speed, Full-Field Deformation Measurements Near Inertial Microcavitation Bubbles Inside Viscoelastic Hydrogels

Contact Info

Product

Resources

About