Traffic safety remains a critical global concern, with timely and accurate accident detection essential for hazard reduction and rapid emergency response. Infrastructure-based vision sensors offer scalable and efficient solutions for continuous real-time monitoring, facilitating automated detection of accidents directly from captured images. This research investigates the zero-shot capabilities of multimodal large language models (MLLMs) for detecting and describing traffic accidents using images from infrastructure cameras, thus minimizing reliance on extensive labeled datasets. Main contributions include: (1) Evaluation of MLLMs using the simulated DeepAccident dataset from CARLA, explicitly addressing the scarcity of diverse, realistic, infrastructure-based accident data through controlled simulations; (2) Comparative performance analysis between Gemini 1.5 and 2.0, Gemma 3 and Pixtral models in accident identification and descriptive capabilities without prior fine-tuning; and (3) Integration of advanced visual analytics, specifically YOLO for object detection, Deep SORT for multi-object tracking, and Segment Anything (SAM) for instance segmentation, into enhanced prompts to improve model accuracy and explainability. Key numerical results show Pixtral as the top performer with an F1-score of 0.71 and 83% recall, while Gemini models gained precision with enhanced prompts (e.g., Gemini 1.5 rose to 90%) but suffered notable F1 and recall losses. Gemma 3 offered the most balanced performance with minimal metric fluctuation. These findings demonstrate the substantial potential of integrating MLLMs with advanced visual analytics techniques, enhancing their applicability in real-world automated traffic monitoring systems.

While most state-of-the-art instance segmentation methods produce binary segmentation masks, geographic and cartographic applications typically require precise vector polygons of extracted objects instead of rasterized output. This paper introduces PolyWorld, a neural network that directly extracts building vertices from an image and connects them correctly to create precise polygons. The model predicts the connection strength between each pair of vertices using a graph neural network and estimates the assignments by solving a differentiable optimal transport problem. Moreover, the vertex positions are optimized by minimizing a combined segmentation and polygonal angle difference loss. PolyWorld significantly outperforms the state of the art in building polygonization and achieves not only notable quantitative results, but also produces visually pleasing building polygons. Code and trained weights are publicly available at this https URL.

Few-shot semantic segmentation is vital for deep learning-based infrastructure inspection applications, where labeled training examples are scarce and expensive. Although existing deep learning frameworks perform well, the need for extensive labeled datasets and the inability to learn new defect categories with little data are problematic. We present our Enhanced Feature Pyramid Network (E-FPN) framework for few-shot semantic segmentation of culvert and sewer defect categories using a prototypical learning framework. Our approach has three main contributions: (1) adaptive E-FPN encoder using InceptionSepConv blocks and depth-wise separable convolutions for efficient multi-scale feature extraction; (2) prototypical learning with masked average pooling for powerful prototype generation from small support examples; and (3) attention-based feature representation through global self-attention, local self-attention and cross-attention. Comprehensive experimentation on challenging infrastructure inspection datasets illustrates that the method achieves excellent few-shot performance, with the best configuration being 8-way 5-shot training configuration at 82.55% F1-score and 72.26% mIoU in 2-way classification testing. The self-attention method had the most significant performance improvements, providing 2.57% F1-score and 2.9% mIoU gain over baselines. Our framework addresses the critical need to rapidly respond to new defect types in infrastructure inspection systems with limited new training data that lead to more efficient and economical maintenance plans for critical infrastructure systems.

Medical image segmentation plays an important role in accurately identifying and isolating regions of interest within medical images. Generative approaches are particularly effective in modeling the statistical properties of segmentation masks that are closely related to the respective structures. In this work we introduce FlowSDF, an image-guided conditional flow matching framework, designed to represent the signed distance function (SDF), and, in turn, to represent an implicit distribution of segmentation masks. The advantage of leveraging the SDF is a more natural distortion when compared to that of binary masks. Through the learning of a vector field associated with the probability path of conditional SDF distributions, our framework enables accurate sampling of segmentation masks and the computation of relevant statistical measures. This probabilistic approach also facilitates the generation of uncertainty maps represented by the variance, thereby supporting enhanced robustness in prediction and further analysis. We qualitatively and quantitatively illustrate competitive performance of the proposed method on a public nuclei and gland segmentation data set, highlighting its utility in medical image segmentation applications.

We investigate the genus $g(n,m)$ of the Erdős-Rényi random graph $G(n,m)$, providing a thorough description of how this relates to the function $m=m(n)$, and finding that there is different behaviour depending on which `region' $m$ falls into. Results already exist for $m \le \frac{n}{2} + O(n^{2/3})$ and $m = \omega \left( n^{1+\frac{1}{j}} \right)$ for $j \in \mathbb{N}$, and so we focus on the intermediate cases. We establish that $g(n,m) = (1+o(1)) \frac{m}{2}$ whp (with high probability) when $n \ll m = n^{1+o(1)}$, that $g(n,m) = (1+o(1)) \mu (\lambda) m$ whp for a given function $\mu (\lambda)$ when $m \sim \lambda n$ for $\lambda > \frac{1}{2}$, and that $g(n,m) = (1+o(1)) \frac{8s^{3}}{3n^{2}}$ whp when $m = \frac{n}{2} + s$ for $n^{2/3} \ll s \ll n$. We then also show that the genus of a fixed graph can increase dramatically if a small number of random edges are added. Given any connected graph with bounded maximum degree, we find that the addition of $\epsilon n$ edges will whp result in a graph with genus $\Omega (n)$, even when $\epsilon$ is an arbitrarily small constant! We thus call this the `fragile genus' property.

It was recently shown that the loss function used for training physics-informed neural networks (PINNs) exhibits local minima at solutions corresponding to fixed points of dynamical systems. In the forward setting, where the PINN is trained to solve initial value problems, these local minima can interfere with training and potentially leading to physically incorrect solutions. Building on stability theory, this paper proposes a regularization scheme that penalizes solutions corresponding to unstable fixed points. Experimental results on four dynamical systems, including the Lotka-Volterra model and the van der Pol oscillator, show that our scheme helps avoiding physically incorrect solutions and substantially improves the training success rate of PINNs.

JWST has identified a large population of faint, broad-line active galactic nuclei (AGN) in the early universe that are powered by black holes (BHs) that often appear overmassive relative to their host galaxies. In this study, we examine the relationship between BH mass and galaxy stellar mass at $33\sigma$ above the relationship measured for local broad-line AGN. We derive an intrinsic scatter in this relationship of $0.9$ dex, which does not vary over the redshift range of our sample. We also find that the $M_{\rm BH}/M_{\star}$ ratio increases by $2.3$ dex from $z = 3.5$ and $z = 6.5$ with a confidence level of > 3\sigma. We attribute this trend with the increasing fraction of LRDs in our sample at z>4 as their host masses are $\sim1$ dex lower than the non-LRD AGN in our sample. These results support a picture in which the BHs powering JWST's broad-line AGN are genuinely overmassive and become increasingly so with redshift. We discuss the implications of our findings on early BH growth relative to that of their host galaxies and the constraints it places on BH seeding models.

Real-time visibility determination in expansive or dynamically changing environments has long posed a significant challenge in computer graphics. Existing techniques are computationally expensive and often applied as a precomputation step on a static scene. We present NeuralPVS, the first deep-learning approach for visibility computation that efficiently determines from-region visibility in a large scene, running at approximately 100 Hz processing with less than $1\%$ missing geometry. This approach is possible by using a neural network operating on a voxelized representation of the scene. The network's performance is achieved by combining sparse convolution with a 3D volume-preserving interleaving for data compression. Moreover, we introduce a novel repulsive visibility loss that can effectively guide the network to converge to the correct data distribution. This loss provides enhanced robustness and generalization to unseen scenes. Our results demonstrate that NeuralPVS outperforms existing methods in terms of both accuracy and efficiency, making it a promising solution for real-time visibility computation.

In this article we propose a novel method for sampling from Gibbs distributions of the form $\pi(x)\propto\exp(-U(x))$ with a potential $U(x)$. In particular, inspired by diffusion models we propose to consider a sequence $(\pi^{t_k})_k$ of approximations of the target density, for which $\pi^{t_k}\approx \pi$ for $k$ small and, on the other hand, $\pi^{t_k}$ exhibits favorable properties for sampling for $k$ large. This sequence is obtained by replacing parts of the potential $U$ by its Moreau envelopes. Sampling is performed in an Annealed Langevin type procedure, that is, sequentially sampling from $\pi^{t_k}$ for decreasing $k$, effectively guiding the samples from a simple starting density to the more complex target. In addition to a theoretical analysis we show experimental results supporting the efficacy of the method in terms of increased convergence speed and applicability to multi-modal densities $\pi$.

Temporal action segmentation in untrimmed videos has gained increased attention recently. However, annotating action classes and frame-wise boundaries is extremely time consuming and cost intensive, especially on large-scale datasets. To address this issue, we propose an unsupervised approach for learning action classes from untrimmed video sequences. In particular, we propose a temporal embedding network that combines relative time prediction, feature reconstruction, and sequence-to-sequence learning, to preserve the spatial layout and sequential nature of the video features. A two-step clustering pipeline on these embedded feature representations then allows us to enforce temporal consistency within, as well as across videos. Based on the identified clusters, we decode the video into coherent temporal segments that correspond to semantically meaningful action classes. Our evaluation on three challenging datasets shows the impact of each component and, furthermore, demonstrates our state-of-the-art unsupervised action segmentation results.