arXiv subjects

Lukun Wang

Publications and source records attributed to Lukun Wang.

3 recordsLinked to original sources

arXiv2025

JoPano: Unified Panorama Generation via Joint Modeling

Panorama generation has recently attracted growing interest in the research community, with two core tasks, text-to-panorama and view-to-panorama generation. However, existing methods still face two major challenges: their U-Net-based architectures constrain the visual quality of the generated panoramas, and they usually treat the two core tasks independently, which leads to modeling redundancy and inefficiency. To overcome these challenges, we propose a joint-face panorama (JoPano) generation approach that unifies the two core tasks within a DiT-based model. To transfer the rich generative capabilities of existing DiT backbones learned from natural images to the panorama domain, we propose a Joint-Face Adapter built on the cubemap representation of panoramas, which enables a pretrained DiT to jointly model and generate different views of a panorama. We further apply Poisson Blending to reduce seam inconsistencies that often appear at the boundaries between cube faces. Correspondingly, we introduce Seam-SSIM and Seam-Sobel metrics to quantitatively evaluate the seam consistency. Moreover, we propose a condition switching mechanism that unifies text-to-panorama and view-to-panorama tasks within a single model. Comprehensive experiments show that JoPano can generate high-quality panoramas for both text-to-panorama and view-to-panorama generation tasks, achieving state-of-the-art performance on FID, CLIP-FID, IS, and CLIP-Score metrics.

cs.CV↗

arXiv2025

DCAMamba: Mamba-based Rapid Response DC Arc Fault Detection

In electrical equipment, even minor contact issues can lead to arc faults. Traditional methods often struggle to balance the accuracy and rapid response required for effective arc fault detection. To address this challenge, we introduce DCAMamba, a novel framework for arc fault detection. Specifically, DCAMamba is built upon a state-space model (SSM) and utilizes a hardware-aware parallel algorithm, designed in a cyclic mode using the Mamba architecture. To meet the dual demands of high accuracy and fast response in arc fault detection, we have refined the original Mamba model and incorporated a Feature Amplification Strategy (FAS), a simple yet effective method that enhances the model's ability to interpret arc fault data. Experimental results show that DCAMamba, with FAS, achieves a 12$\%$ improvement in accuracy over the original Mamba, while maintaining an inference time of only 1.87 milliseconds. These results highlight the significant potential of DCAMamba as a future backbone for signal processing. Our code will be made open-source after peer review.

cs.CE↗

arXiv2024

StyleBrush: Style Extraction and Transfer from a Single Image

Stylization for visual content aims to add specific style patterns at the pixel level while preserving the original structural features. Compared with using predefined styles, stylization guided by reference style images is more challenging, where the main difficulty is to effectively separate style from structural elements. In this paper, we propose StyleBrush, a method that accurately captures styles from a reference image and ``brushes'' the extracted style onto other input visual content. Specifically, our architecture consists of two branches: ReferenceNet, which extracts style from the reference image, and Structure Guider, which extracts structural features from the input image, thus enabling image-guided stylization. We utilize LLM and T2I models to create a dataset comprising 100K high-quality style images, encompassing a diverse range of styles and contents with high aesthetic score. To construct training pairs, we crop different regions of the same training image. Experiments show that our approach achieves state-of-the-art results through both qualitative and quantitative analyses. We will release our code and dataset upon acceptance of the paper.

cs.CV↗