arXiv subjects

Shashikant Verma

Publications and source records attributed to Shashikant Verma.

9 recordsLinked to original sources

arXiv2026

Dynamics of an internally actuated elastic particle in a plane Poiseuille flow

We analytically analyse the dynamics of an internally actuated particle, modelled as a compressible elastic sphere embedded with a magnetic bead at its undeformed centre, translating in a plane Poiseuille flow in the Stokes limit. The particle is constrained to translate with a prescribed velocity while remaining at an arbitrary position within the flow by applying an external point force and external point torque at its undeformed centre. The governing equations for the fluid and particle are the Stokes and Navier elasticity equations, respectively. We use the series solutions to the governing equations and the domain perturbation method to capture the deformed shape of the particle, assuming $\alpha \ll 1$. Here, $\alpha$ quantifies the elastic strain induced in the particle due to the viscous stress from the fluid. The external force and external torque are obtained until O($\alpha^2$). The particle translating along the channel length experiences an elastic-induced hydrodynamic lift as well as hydrodynamic torque both at O($\alpha$) and O($\alpha^2$). The leading-order lift depends linearly on the local shear rate and on the combined effects of slip velocity and flow curvature, where the slip velocity is defined as the particle velocity relative to the local ambient flow. The particle reaches a stable equilibrium position away from the centreline, where the net lift vanishes. We show that the direction of deformation-induced lateral migration of the internally actuated particle is qualitatively distinct from that of drops, capsules, and vesicles in the Stokes limit and from that of rigid spheres undergoing inertial migration.

physics.flu-dyn↗

arXiv2025

Dynamics of an internally actuated weakly elastic sphere in a general quadratic flow

Internally actuated elastic particles are widely used in biomedical applications. It is imperative to understand the dynamics of such particles in pressure-driven microfluidic devices to manipulate their motion. We analytically examine the dynamics of an internally actuated elastic particle translating in a general unbounded quadratic flow in the inertialess limit. We consider the particle as a compressible weakly elastic sphere, and its motion is controlled by applying an external point force and a point torque at the centre of its undeformed shape. The fluid and the particle are modelled using the Stokes and the Navier elasticity equations, respectively. We use the domain perturbation method to capture the particle deformation. The point force and the point torque are obtained until \textit{O}($\alpha^2$), assuming $\alpha\ll 1$. Here, $\alpha$ is the measure of the particle elastic strain induced due to the fluid viscous stress. We present the results for the particle motion in a general unbounded quadratic flow. The results are simplified further for the motion along the centreline in the quadratic component of three Poiseuille flows: 1) elliptical Poiseuille, 2) plane Poiseuille, and 3) Hagen-Poiseuille flows. In the general quadratic flow, the point force at \textit{O}($\alpha$) is aligned with the particle velocity, while the force at \textit{O}($\alpha^2$) acts at an angle to the velocity. Furthermore, the torque is non-zero due to elastic effects at \textit{O}($\alpha$) and \textit{O}($\alpha^2$). For all the three Poiseuille flows, the point force until \textit{O}($\alpha^2$) is aligned with the particle velocity, while the torque comes as zero.

physics.flu-dyn↗

arXiv2025

UnCageNet: Tracking and Pose Estimation of Caged Animal

Animal tracking and pose estimation systems, such as STEP (Simultaneous Tracking and Pose Estimation) and ViTPose, experience substantial performance drops when processing images and videos with cage structures and systematic occlusions. We present a three-stage preprocessing pipeline that addresses this limitation through: (1) cage segmentation using a Gabor-enhanced ResNet-UNet architecture with tunable orientation filters, (2) cage inpainting using CRFill for content-aware reconstruction of occluded regions, and (3) evaluation of pose estimation and tracking on the uncaged frames. Our Gabor-enhanced segmentation model leverages orientation-aware features with 72 directional kernels to accurately identify and segment cage structures that severely impair the performance of existing methods. Experimental validation demonstrates that removing cage occlusions through our pipeline enables pose estimation and tracking performance comparable to that in environments without occlusions. We also observe significant improvements in keypoint detection accuracy and trajectory consistency.

cs.CV↗

arXiv2025

PanoHair: Detailed Hair Strand Synthesis on Volumetric Heads

Achieving realistic hair strand synthesis is essential for creating lifelike digital humans, but producing high-fidelity hair strand geometry remains a significant challenge. Existing methods require a complex setup for data acquisition, involving multi-view images captured in constrained studio environments. Additionally, these methods have longer hair volume estimation and strand synthesis times, which hinder efficiency. We introduce PanoHair, a model that estimates head geometry as signed distance fields using knowledge distillation from a pre-trained generative teacher model for head synthesis. Our approach enables the prediction of semantic segmentation masks and 3D orientations specifically for the hair region of the estimated geometry. Our method is generative and can generate diverse hairstyles with latent space manipulations. For real images, our approach involves an inversion process to infer latent codes and produces visually appealing hair strands, offering a streamlined alternative to complex multi-view data acquisition setups. Given the latent code, PanoHair generates a clean manifold mesh for the hair region in under 5 seconds, along with semantic and orientation maps, marking a significant improvement over existing methods, as demonstrated in our experiments.

cs.GR↗

arXiv2025

SemFaceEdit: Semantic Face Editing on Generative Radiance Manifolds

Despite multiple view consistency offered by 3D-aware GAN techniques, the resulting images often lack the capacity for localized editing. In response, generative radiance manifolds emerge as an efficient approach for constrained point sampling within volumes, effectively reducing computational demands and enabling the learning of fine details. This work introduces SemFaceEdit, a novel method that streamlines the appearance and geometric editing process by generating semantic fields on generative radiance manifolds. Utilizing latent codes, our method effectively disentangles the geometry and appearance associated with different facial semantics within the generated image. In contrast to existing methods that can change the appearance of the entire radiance field, our method enables the precise editing of particular facial semantics while preserving the integrity of other regions. Our network comprises two key modules: the Geometry module, which generates semantic radiance and occupancy fields, and the Appearance module, which is responsible for predicting RGB radiance. We jointly train both modules in adversarial settings to learn semantic-aware geometry and appearance descriptors. The appearance descriptors are then conditioned on their respective semantic latent codes by the Appearance Module, facilitating disentanglement and enhanced control. Our experiments highlight SemFaceEdit's superior performance in semantic field-based editing, particularly in achieving improved radiance field disentanglement.

cs.CV↗

arXiv2025

DMD-Net: Deep Mesh Denoising Network

We present Deep Mesh Denoising Network (DMD-Net), an end-to-end deep learning framework, for solving the mesh denoising problem. DMD-Net consists of a Graph Convolutional Neural Network in which aggregation is performed in both the primal as well as the dual graph. This is realized in the form of an asymmetric two-stream network, which contains a primal-dual fusion block that enables communication between the primal-stream and the dual-stream. We develop a Feature Guided Transformer (FGT) paradigm, which consists of a feature extractor, a transformer, and a denoiser. The feature extractor estimates the local features, that guide the transformer to compute a transformation, which is applied to the noisy input mesh to obtain a useful intermediate representation. This is further processed by the denoiser to obtain the denoised mesh. Our network is trained on a large scale dataset of 3D objects. We perform exhaustive ablation studies to demonstrate that each component in our network is essential for obtaining the best performance. We show that our method obtains competitive or better results when compared with the state-of-the-art mesh denoising algorithms. We demonstrate that our method is robust to various kinds of noise. We observe that even in the presence of extremely high noise, our method achieves excellent performance.

cs.CV↗

arXiv2025

Dynamics of an internally actuated weakly elastic sphere translating parallel to a rigid wall

We analyse the dynamics of a weakly elastic spherical particle translating parallel to a rigid wall in a quiescent Newtonian fluid in the Stokes limit. The particle motion is constrained parallel to the wall by applying a point force and a point torque at the centre of its undeformed shape. The particle is modelled using the Navier elasticity equations. The series solutions to the Navier and the Stokes equations are utilised to obtain the displacement and velocity fields in the solid and fluid, respectively. The point force and the point torque are calculated as series in small parameters $\alpha$ and $1/H$, using the domain perturbation method and the method of reflections. Here, $\alpha$ is the measure of elastic strain induced in the particle resulting from the fluid's viscous stress, and $H$ is the non-dimensional gap width, defined as the ratio of the distance of the particle centre from the wall to its radius. The results are presented up to $\textit{O}(1/H^3)$ and $\textit{O}(1/H^2)$, assuming $\alpha \sim 1/H$, for cases where gravity is aligned and non-aligned with the particle velocity, respectively. The deformed shape of the particle is determined by the force distribution acting on it. The hydrodynamic lift due to elastic effects (acting away from the wall) appears at $\textit{O}(\alpha/H^2)$, in the former case. In an unbounded domain, the elastic effects in the latter case generate a hydrodynamic torque at \textit{O}($\alpha$) and a drag at \textit{O}($\alpha^2$). Conversely, in the former case, the torque is zero, while the drag still appears at \textit{O}($\alpha^2$).

physics.flu-dyn↗

arXiv2025

STEP: Simultaneous Tracking and Estimation of Pose for Animals and Humans

We introduce STEP, a novel framework utilizing Transformer-based discriminative model prediction for simultaneous tracking and estimation of pose across diverse animal species and humans. We are inspired by the fact that the human brain exploits spatiotemporal continuity and performs concurrent localization and pose estimation despite the specialization of brain areas for form and motion processing. Traditional discriminative models typically require predefined target states for determining model weights, a challenge we address through Gaussian Map Soft Prediction (GMSP) and Offset Map Regression Adapter (OMRA) Modules. These modules remove the necessity of keypoint target states as input, streamlining the process. Our method starts with a known target state in the initial frame of a given video sequence. It then seamlessly tracks the target and estimates keypoints of anatomical importance as output for subsequent frames. Unlike prevalent top-down pose estimation methods, our approach doesn't rely on per-frame target detections due to its tracking capability. This facilitates a significant advancement in inference efficiency and potential applications. We train and validate our approach on datasets encompassing diverse species. Our experiments demonstrate superior results compared to existing methods, opening doors to various applications, including but not limited to action recognition and behavioral analysis.

cs.CV↗

arXiv2025

L3D-Pose: Lifting Pose for 3D Avatars from a Single Camera in the Wild

While 2D pose estimation has advanced our ability to interpret body movements in animals and primates, it is limited by the lack of depth information, constraining its application range. 3D pose estimation provides a more comprehensive solution by incorporating spatial depth, yet creating extensive 3D pose datasets for animals is challenging due to their dynamic and unpredictable behaviours in natural settings. To address this, we propose a hybrid approach that utilizes rigged avatars and the pipeline to generate synthetic datasets to acquire the necessary 3D annotations for training. Our method introduces a simple attention-based MLP network for converting 2D poses to 3D, designed to be independent of the input image to ensure scalability for poses in natural environments. Additionally, we identify that existing anatomical keypoint detectors are insufficient for accurate pose retargeting onto arbitrary avatars. To overcome this, we present a lookup table based on a deep pose estimation method using a synthetic collection of diverse actions rigged avatars perform. Our experiments demonstrate the effectiveness and efficiency of this lookup table-based retargeting approach. Overall, we propose a comprehensive framework with systematically synthesized datasets for lifting poses from 2D to 3D and then utilize this to re-target motion from wild settings onto arbitrary avatars.

cs.CV↗