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.
In this paper, we work with simple and finite graphs. We study a generalization of the \emph{Cage Problem}, which has been widely studied since cages were introduced by Tutte \cite{T47} in 1947 and after Erd\" os and Sachs \cite{ES63} proved their existence in 1963. An \emph{$(r,g)$-graph} is an $r$-regular graph in which the shortest cycle has length equal to $g$; that is, it is an $r$-regular graph with girth $g$. An \emph{$(r,g)$-cage} is an $(r,g)$-graph with the smallest possible number of vertices among all $(r,g)$-graphs; the order of an $(r,g)$-cage is denoted by $n(r,g)$. The Cage Problem consists of finding $(r,g)$-cages; it is well-known that $(r,g)$-cages have been determined only for very limited sets of parameter pairs $(r, g)$. There exists a simple lower bound for $n(r,g)$, given by Moore and denoted by $n_0(r,g)$. The cages that attain this bound are called \emph{Moore cages}.
As 3D Gaussian Splatting (3DGS) gains popularity as a 3D representation of real scenes, enabling user-friendly deformation to create novel scenes while preserving fine details from the original 3DGS has attracted significant research attention. We introduce CAGE-GS, a cage-based 3DGS deformation method that seamlessly aligns a source 3DGS scene with a user-defined target shape. Our approach learns a deformation cage from the target, which guides the geometric transformation of the source scene. While the cages effectively control structural alignment, preserving the textural appearance of 3DGS remains challenging due to the complexity of covariance parameters. To address this, we employ a Jacobian matrix-based strategy to update the covariance parameters of each Gaussian, ensuring texture fidelity post-deformation. Our method is highly flexible, accommodating various target shape representations, including texts, images, point clouds, meshes and 3DGS models. Extensive experiments and ablation studies on both public datasets and newly proposed scenes demonstrate that our method significantly outperforms existing techniques in both efficiency and deformation quality.
Generalization in robotic manipulation remains a critical challenge, particularly when scaling to new environments with limited demonstrations. This paper introduces CAGE, a novel robotic manipulation policy designed to overcome these generalization barriers by integrating a causal attention mechanism. CAGE utilizes the powerful feature extraction capabilities of the vision foundation model DINOv2, combined with LoRA fine-tuning for robust environment understanding. The policy further employs a causal Perceiver for effective token compression and a diffusion-based action prediction head with attention mechanisms to enhance task-specific fine-grained conditioning. With as few as 50 demonstrations from a single training environment, CAGE achieves robust generalization across diverse visual changes in objects, backgrounds, and viewpoints. Extensive experiments validate that CAGE significantly outperforms existing state-of-the-art RGB/RGB-D approaches in various manipulation tasks, especially under large distribution shifts. In similar environments, CAGE offers an average of 42% increase in task completion rate. While all baselines fail to execute the task in unseen environments, CAGE
The non-Hermitian Aharonov-Bohm (AB) cage is a unique localization phenomenon that confines all possible excitations. This confinement leads to fully flat spectra in momentum space, which are typically accompanied with the degeneracy with various types. Classifying the degeneracy type is crucial for studying the dynamical properties of the non-Hermitian AB cage, but the methods for such classification and their physical connections remain not very clear. Here, we construct a non-Hermitian AB cage in a bosonic Bogoliubov-de Gennes (BdG) system with various types of degenerate flat bands (DFBs). Using the transfer matrix, we demonstrate the localization mechanism for the formation of AB cage and derive the minimal polynomial in mathematics for classifying the degeneracy types of DFBs, thus providing comprehensive understanding of the correspondence among the degeneracy type of DFBs, the minimal polynomial, and the transfer matrix. With such correspondence, we propose a scheme to realize highly degenerate flat bands.
Active matter can flow and yield under conditions where passive matter jams and slows down, as self-propulsion significantly modulates particle escape from local cages. How activity microscopically reshapes the caging environment to produce this effect, however, remains poorly understood. Here we study a minimal active-matter model of cage breaking: three distinguishable self-propelling disks under circular confinement. This simple setting allows us to construct an entropic landscape for rearrangements and to compare it exactly with its equilibrium counterpart. At low activity the landscape is effectively bistable, whereas at high activity it develops additional metastable basins associated with frustrated clusters at the boundary. We quantify the system's departure from equilibrium and show that cage breaking is fastest when the persistence length matches the particle radius, linking a geometric microscopic scale to the enhanced dynamics of active glasses. Extending the landscape to two dimensions reveals circulating probability currents, and a Markov-state description shows that detailed balance is broken both in the continuous landscape dynamics and in the coarse-grained transit
Understanding the role that structure plays in the dynamical arrest observed in glassy systems remains an open challenge. Over the last decade, machine learning (ML) strategies have emerged as an important tool for probing this structure-dynamics relationship, particularly for predicting heterogeneous glassy dynamics from local structure. A recent advancement is the introduction of the cage state, a structural quantity that captures the average positions of particles while rearrangements are forbidden. During the caging regime, linear models trained on the cage state have been shown to outperform more complex ML methods trained on initial configurations only. In this paper, we explore the properties associated with the cage state in more detail to better understand why it serves as such an effective predictor for the dynamics. Specifically, we examine how the cage state in a binary hard-sphere mixture is influenced by both packing fraction and boundary conditions. Our results reveal that, as the system approaches the glassy regime, the cage state becomes increasingly influenced by long-range structural effects. This influence is evident both in its predictive power for particle dyn
A graph $G$ is cyclically $c$-edge-connected if there is no set of fewer than $c$ edges that disconnects $G$ into at least two cyclic components. We prove that if a $(k, g)$-cage $G$ has at most $2M(k, g) - g^2$ vertices, where $M(k, g)$ is the Moore bound, then $G$ is cyclically $(k - 2)g$-edge-connected, which equals the number of edges separating a $g$-cycle, and every cycle-separating $(k - 2)g$-edge-cut in $G$ separates a cycle of length $g$. In particular, this is true for unknown cages with $(k, g) \in \{(3, 13), (3, 14), (3, 15), (4, 9), (4, 10)$, $(4, 11),$ $(5, 7), (5, 9), (5, 10), (5, 11), (6, 7), (9, 7)\}$ and also the potential missing Moore graph with degree $57$ and diameter $2$. Keywords: cage, cyclic connectivity, girth, lower bound
We experimentally explore the morphological evolution of cages in quasi-two-dimensional suspensions of colloidal fluids, uncovering a complex dynamic restructuring in the fluid. Although cages display isotropic evolution in the laboratory frame, we observe a striking anisotropy when analyzed in the displacement frame of the caged particles. Moreover, our findings reveal that particles in specific but distinct regions of the cage predominantly contribute to either its persistence or relaxation. Thus, our study provides a coarse-grained microscopic picture of the structural relaxation of these fluids through cage evolution, which has broader implications for the flow and phase behavior of complex fluids in confined geometry.
We address the question of how to connect predictions by hydrodynamic models of how sea lice move in water to observable measures that count the number of lice on each fish in a cage in the water. This question is important for management and regulation of aquacultural practice that tries to maximise food production and minimise risk to the environment. We do this through a simple rule-based model of interaction between sea lice and caged fish. The model is simple: sea lice can attach and detach from a fish. The model has a novel feature, encoding what is known as a master equation producing a time-series of distributions of lice on fish that one might expect to find if a cage full of fish were placed at any given location. To demonstrate how this works, and to arrive at a rough estimate of the interaction rates, we fit a simplified version of the model with three free parameters to publicly available data about an experiment with sentinel cages in Loch Linnhe in Scotland. Our construction, coupled to the hydrodynamic models driven by surveillance data from industrial farms, quantifies the environmental impact as: what would the infection burden look like in a notional cage at any
WebAssembly (WASM) is an immensely versatile and increasingly popular compilation target. It executes applications written in several languages (e.g., C/C++) with near-native performance in various domains (e.g., mobile, edge, cloud). Despite WASM's sandboxing feature, which isolates applications from other instances and the host platform, WASM does not inherently provide any memory safety guarantees for applications written in low-level, unsafe languages. To this end, we propose Cage, a hardware-accelerated toolchain for WASM that supports unmodified applications compiled to WASM and utilizes diverse Arm hardware features aiming to enrich the memory safety properties of WASM. Precisely, Cage leverages Arm's Memory Tagging Extension (MTE) to (i) provide spatial and temporal memory safety for heap and stack allocations and (ii) improve the performance of WASM's sandboxing mechanism. Cage further employs Arm's Pointer Authentication (PAC) to prevent leaked pointers from being reused by other WASM instances, thus enhancing WASM's security properties. We implement our system based on 64-bit WASM. We provide a WASM compiler and runtime with support for Arm's MTE and PAC. On top of that,
Let $2 \le r < m$ and $g$ be positive integers. An $({r,m};g)$--graph} (or biregular graph) is a graph with degree set ${r,m}$ and girth $g$, and an $({r,m};g)$-cage (or biregular cage) is an $({r,m};g)$-graph of minimum order $n({r,m};g)$. If $m=r+1$, an $({r,m};g)$-cage is said to be a semiregular cage. In this paper we generalize the reduction and graph amalgam operations from M. Abreu, G. Araujo-Pardo, C. Balbuena, D. Labbate (2011) on the incidence graphs of an affine and a biaffine plane obtaining two new infinite families of biregular cages and two new semiregular cages. The constructed new families are $({r,2r-3};5)$-cages for all $r=q+1$ with $q$ a prime power, and $({r,2r-5};5)$-cages for all $r=q+1$ with $q$ a prime. The new semiregular cages are constructed for r=5 and 6 with 31 and 43 vertices respectively.
A [z,r;g]-mixed cage is a mixed graph of minimum order such that each vertex has z in-arcs, z out-arcs, r edges, and it has girth g. We present an infinite family of mixed graphs with girth 6. This construction also provides an upper bound on the minimum order of mixed cages of girth 6. Additionally,we introduce a lower bound on the minimum order for any mixed cage.
A novel growth technique for low density, non-icosahedral carbon onion-like structures on Cu surface is described. The technique differs with the formation of carbon onions inside the C1+ implanted metal surfaces at high temperatures with densities ~ 2 g cm-3 in the form of multishelled icosahedral fullerenes wrapped around C60. We report spheroidal cage formation of shell with variable shell thickness, radii and curvature on C1+-irradiated Cu surfaces by the accumulating C atoms which emerge out of the edges of the implanted sheets and grids. Atom-by-atom C accreting structures grow along the edges, corners and crevices. These, in turn, form the open and closed multiple-shelled cages. Variable curvature is the most common feature of the growing structures. The growth at room temperature does not ensure sphericity of the cages due to the absence of high temperature annealing. These cages have very low density, hollow interiors and often exhibit the conspicuous cage-inside-cage structures. Another feature of the as-grown, non-spherical onions is the existence of spherical protrusions with different local radii within the same cage. Formation of cages with sizes ~ few nanometer to hu
We study transport of interacting bosons through an Aharonov-Bohm cage - a building block of flat band networks - with coherent pump and sink leads. In the absence of interactions the cage is insulating due to destructive interference. We find that the cage stays insulating up to a critical value of the pump strength in the presence of mean field interactions, while the quantum regime induces particle pair transport and weak conductance below the critical pump strength. A swift crossover from quantum into the classical regime upon further pump strength increase is observed. We solve the time dependent master equations for the density matrix of the many body problem both in the classical, pure quantum, and pseudoclassical regimes. We start with an empty cage and switch on driving. We characterize the transient dynamics, and the complexity of the resulting steady states and attractors. Our results can be readily realized using experimental platforms involving interacting ultracold atoms and photons on finetuned optical lattices.
A study of the wave dynamics around a multiple fishing cage system is carried out under the assumption of the linear water wave theory and small-amplitude wave response. The Fourier--Bessel series expansion of the velocity potential is derived for regions enclosed under the open-water and cage systems and in the immediate vicinity. Further, the scattering between the cages is accounted for by employing Graf's addition theorem. The porous flexible cage system is modelled using Darcy's law and the three-dimensional membrane equation. The edges of the cages are moored along its circumferences to balance its position in the deep sea. The unknown coefficients in the potentials are obtained by employing the matched eigenfunction method in conjunction with the least-squares approximation method. In addition, the far-field scattering coefficients for the entire system are obtained by expanding the Bessel and Hankel functions in the plane wave representation form. Numerical results such as the hydrodynamic forces, scattering coefficients, and power dissipation are investigated for various cage and wave parameters. The wave loading on the cage system can be significantly damped by the spatia
The relationship between structure and dynamics in glassy fluids remains an intriguing open question. Recent work has shown impressive advances in our ability to predict local dynamics using structural features, most notably due to the use of advanced machine learning techniques. Here we explore whether a simple linear regression algorithm combined with intelligently chosen structural order parameters can reach the accuracy of the current, most advanced machine learning approaches for predicting dynamic propensity. To do this we introduce a method to pinpoint the cage state of the initial configuration -- i.e. the configuration consisting of the average particle positions when particle rearrangement is forbidden. We find that, in comparison to both the initial state and the inherent state, the structure of the cage state is highly predictive of the long-time dynamics of the system. Moreover, by combining the cage state information with the initial state, we are able to predict dynamic propensities with unprecedentedly high accuracy over a broad regime of time scales, including the caging regime.
Glassy dynamics is intermittent, as particles suddenly jump out of the cage formed by their neighbours, and heterogeneous, as these jumps are not uniformly distributed across the system. Relating these features of the dynamics to the diverse local environments explored by the particles is essential to rationalize the relaxation process. Here we investigate this issue characterizing the local environment of a particle with the amplitude of its short time vibrational motion, as determined by segmenting in cages and jumps the particle trajectories. Both simulations of supercooled liquids and experiments on colloidal suspensions show that particles in large cages are likely to jump after a small time-lag, and that, on average, the cage enlarges shortly before the particle jumps. At large time-lags, the cage has essentially a constant value, which is smaller for longer-lasting cages. Finally, we clarify how this coupling between cage size and duration controls the average behaviour and opens the way to a better understanding of the relaxation process in glass--forming liquids.
We examine the vibrational stability of three candidate structures for the B24N24 cage and report their infra-red (IR) and Raman spectra. The candidate structures considered are a round cage with octahedral O symmetry, a cage with S_4 symmetry that satisfies the isolated square rule, and a cage of S_8 symmetry, which combines the caps of the (4,4) nanotube, and contains two extra squares and octagons. The calculations are performed within density functional theory, at the all electron level, with large basis sets, and within the generalized gradient approximation. The vertical ionization potential (VIP) and static dipole polarizability are also reported. The S_4 and S_8 cages are energetically nearly degenerate and are favored over the O cage which has six extra octagons and squares. The IR and Raman spectra of the three clusters show notable differences providing thereby a way to identify and possibly synthesize the cages.
The clathrate-like or caged compounds have attracted continuing interest due to their structural flexibility as well as fertile physical properties. Here we report a pressure-induced reemergence of superconductivity in BaIr2Ge7 and Ba3Ir4Ge16, two new caged superconductors with two-dimensional building blocks of cage structures. After suppressing the ambient-pressure superconducting (SC-I) state, a new superconducting (SC-II) state emerges unexpectedly, with Tc increased to a maximum of 4.4 K and 4.0 K for BaIr2Ge7 and Ba3Ir4Ge16, respectively. Combined with high-pressure synchrotron x-ray diffraction and Raman measurements, we propose that the reemergence of superconductivity in the caged superconductors can be ascribed to a pressure-induced phonon softening linked to the cage shrink.