检索结果-内蒙古大学图书馆

Hybrid Deep Learning Architectures for Histological Image Segmentation 6

学校读者我要写书评

暂无评论

Hybrid Deep Learning Architectures for Histological Image Se...

6th International Conference on Artificial Intelligence in Information and Communication, ICAIIC 2024

作者： Hasan, Md Jahid Ahmad, Wan Siti Halimatul Munirah Wan Ahmad Fauzi, Mohammad Faizal Abas, Fazly Salleh Multimedia University Faculty of Engineering Cyberjaya Malaysia Multimedia University Center for Visual Computing Faculty of Computing and InformaticsFaculty of Engineering Cyberjaya Malaysia Multimedia University Faculty of Engineering and Technology Cyberjaya Malaysia

ISBN: (纸本)9798350344349

In histopathology image analysis, accurate segmentation of nuclei holds immense significance, particularly in the early detection and treatment of diseases like breast cancer. Nuclei segmentation is a fundamental but challenging task because of the intricate variations in nuclear shapes, sizes, densities, and overlapping instances. In this paper, we evaluate eight convolutional neural network (CNN) models, two of them existing models namely U-Net, SegNet, and six hybrid models by combining U-Net and SegNet modify decoder with ResNet, VGG and DenseNet (ResNet-UNet, ResNet-SegNet, VGG-UNet, VGG-SegNet, DenseNet-UNet, and DenseNet-SegNet. This experiment aims to identify the best deep-learning model for segmenting hematoxylin and eosin (HandE) stain images using a publicly available dataset called MoNuSeg. From the experimented work, we found that VGG-UNet outperforms other models with an F1 score of 0.8452 and IoU of 0.6929 respectively. This research will serve as a foundation for the future construction of more complex deep learning models with cascade or any combination of the models studied. © 2024 IEEE.

关键词： Image segmentation

Deformable One-Shot Face Stylization via DINO Semantic Guidance

学校读者我要写书评

暂无评论

Deformable One-Shot Face Stylization via DINO Semantic Guida...

Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Yang Zhou Zichong Chen Hui Huang Visual Computing Research Center Shenzhen University

ISBN: (数字)9798350353006

ISBN: (纸本)9798350353013

This paper addresses the complex issue of one-shot face stylization, focusing on the simultaneous consideration of appearance and structure, where previous methods have fallen short. We explore deformation-aware face stylization that diverges from traditional single-image style reference, opting for a real-style image pair instead. The cornerstone of our method is the utilization of a self-supervised vision transformer, specifically DINO-ViT, to establish a robust and consistent facial structure representation across both real and style domains. Our stylization process begins by adapting the StyleGAN generator to be deformation-aware through the integration of spatial transformers (STN). We then introduce two innovative constraints for generator fine-tuning under the guidance of DINO semantics: i) a directional deformation loss that regulates directional vectors in DINO space, and ii) a relative structural consistency constraint based on DINO token self-similarities, ensuring diverse generation. Additionally, style-mixing is employed to align the color generation with the reference, minimizing inconsistent correspondences. This framework delivers enhanced deformability for general one-shot face stylization, achieving notable efficiency with a fine-tuning duration of approximately 10 minutes. Extensive qualitative and quantitative comparisons demonstrate our superiority over state-of-the-art one-shot face stylization methods. Code is available at https://***/zichongc/DoesFS.

关键词： Computer vision Codes Deformation Image color analysis Face recognition Semantics Focusing

WindPoly: Polygonal Mesh Reconstruction via Winding Numbers

学校读者我要写书评

暂无评论

arXiv 2024年

作者： He, Xin Lv, Chenlei Huang, Pengdi Huang, Hui Visual Computing Research Center Shenzhen University China

Polygonal mesh reconstruction of a raw point cloud is a valuable topic in the field of computer graphics and 3D vision. Especially to 3D architectural models, polygonal mesh provides concise expressions for fundamental geometric structures while effectively reducing data volume. However, there are some limitations of traditional reconstruction methods: normal vector dependency, noisy points and defective parts sensitivity, and internal geometric structure lost, which reduce the practicality in real scene. In this paper, we propose a robust and efficient polygonal mesh reconstruction method to address the issues in architectural point cloud reconstruction task. It is an iterative adaptation process to detect planar shapes from scattered points. The initial structural polygonal mesh can be established in the constructed convex polyhedral space without assistance of normal vectors. Then, we develop an efficient polygon-based winding number strategy to orient polygonal mesh with global consistency. The significant advantage of our method is to provide a structural reconstruction for architectural point clouds and avoid point-based normal vector analysis. It effectively improves the robustness to noisy points and defective parts. More geometric details can be preserved in the reconstructed polygonal mesh. Experimental results show that our method can reconstruct concise, oriented and faithfully polygonal mesh that are better than results of state-of-the-art methods. More results and details can be found on https://***/research/2024/WindPoly. Copyright © 2024, The Authors. All rights reserved.

关键词： Mesh generation

Generating 3D House Wireframes with Semantics

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Ma, Xueqi Liu, Yilin Zhou, Wenjun Wang, Ruowei Huang, Hui Visual Computing Research Center Shenzhen University China

We present a new approach for generating 3D house wireframes with semantic enrichment using an autoregressive model. Unlike conventional generative models that independently process vertices, edges, and faces, our approach employs a unified wire-based representation for improved coherence in learning 3D wireframe structures. By re-ordering wire sequences based on semantic meanings, we facilitate seamless semantic integration during sequence generation. Our two-phase technique merges a graph-based autoencoder with a transformer-based decoder to learn latent geometric tokens and generate semantic-aware wireframes. Through iterative prediction and decoding during inference, our model produces detailed wireframes that can be easily segmented into distinct components, such as walls, roofs, and rooms, reflecting the semantic essence of the shape. Empirical results on a comprehensive house dataset validate the superior accuracy, novelty, and semantic fidelity of our model compared to existing generative models. More results and details can be found on https://***/research/2024/3DWire. © 2024, CC BY-NC-ND.

关键词： Semantics

Deformable One-shot Face Stylization via DINO Semantic Guidance

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhou, Yang Chen, Zichong Huang, Hui Visual Computing Research Center Shenzhen University China

关键词： Semantics

Introducing Unbiased Depth into 2D Gaussian Splatting for High-accuracy Surface Reconstruction

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Peng, Xiaoming Yang, Yixin Zhou, Yang Huang, Hui Visual Computing Research Center CCSE Shenzhen University China

Recently, 2D Gaussian Splatting (2DGS) has demonstrated superior geometry reconstruction quality than the popular 3DGS by using 2D surfels to approximate thin surfaces. However, it falls short when dealing with glossy surfacess, resulting in visible holes in these areas. We found the reflection discontinuity causes the issue. To fit the jump from diffuse to specular reflection at different viewing angles, depth bias is introduced in the optimized Gaussian primitives. To address that, we first replace the depth distortion loss in 2DGS with a novel depth convergence loss, which imposes a strong constraint on depth continuity. Then, we rectified the depth criterion in determining the actual surface, which fully accounts for all the intersecting Gaussians along the ray. Qualitative and quantitative evaluations across various datasets reveal that our method significantly improves reconstruction quality, with more complete and accurate surfaces than 2DGS. Copyright © 2025, The Authors. All rights reserved.

关键词： Surface reconstruction

Attention Distillation: A Unified Approach to visual Characteristics Transfer

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zhou, Yang Gao, Xu Chen, Zichong Huang, Hui Visual Computing Research Center CSSE Shenzhen University China

Recent advances in generative diffusion models have shown a notable inherent understanding of image style and semantics. In this paper, we leverage the self-attention features from pretrained diffusion networks to transfer the visual characteristics from a reference to generated images. Unlike previous work that uses these features as plug-and-play attributes, we propose a novel attention distillation loss calculated between the ideal and current stylization results, based on which we optimize the synthesized image via backpropagation in latent space. Next, we propose an improved Classifier Guidance that integrates attention distillation loss into the denoising sampling process, further accelerating the synthesis and enabling a broad range of image generation applications. Extensive experiments have demonstrated the extraordinary performance of our approach in transferring the examples’ style, appearance, and texture to new images in synthesis. Code is available at https://***/xugao97/ AttentionDistillation. Copyright © 2025, The Authors. All rights reserved.

关键词： Image denoising

Mixture of Experts for Audio-visual Learning 38

学校读者我要写书评

暂无评论

Mixture of Experts for Audio-Visual Learning

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Cheng, Ying Li, Yang He, Junjie Feng, Rui School of Computer Science Fudan University China Shanghai Key Laboratory of Intelligent Information Processing China Shanghai Collaborative Innovation Center of Intelligent Visual Computing China

With the rapid development of multimedia technology, audio-visual learning has emerged as a promising research topic within the field of multimodal analysis. In this paper, we explore parameter-efficient transfer learning for audio-visual learning and propose the Audio-visual Mixture of Experts (AVMoE) to inject adapters into pre-trained models flexibly. Specifically, we introduce unimodal and cross-modal adapters as multiple experts to specialize in intra-modal and inter-modal information, respectively, and employ a lightweight router to dynamically allocate the weights of each expert according to the specific demands of each task. Extensive experiments demonstrate that our proposed approach AVMoE achieves superior performance across multiple audio-visual tasks, including AVE, AVVP, AVS, and AVQA. Furthermore, visual-only experimental results also indicate that our approach can tackle challenging scenes where modality information is missing. The source code is available at https://***/yingchengy/AVMOE. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

Silo: Half-Gigapixel Cylindrical Stereoscopic Immersive Display

学校读者我要写书评

暂无评论

Silo: Half-Gigapixel Cylindrical Stereoscopic Immersive Disp...

IEEE Annual International Symposium Virtual Reality

作者： Saeed Boorboor Doris Gutiérrez-Rosales Ahamed Shoaib Chahat Kalsi Yue Wang Yuyang Cao Xianfeng Gu Arie E. Kaufman Center for Visual Computing Stony Brook University Stony Brook NY

ISBN: (数字)9798331536459

ISBN: (纸本)9798331536466

We present the design and construction of the Silo, a fully immersive stereoscopic cylindrical tiled-display visualization facility. Comprising 168 high-density LCD displays, the facility provides an ultra-high-resolution image of 619 million pixels, and close to 360 horizontal field-of-regards (FoR), aiming to maximize visual acuity and completely engage the human visual sensorium and its periphery. In this article, we outline the motivations, design principles, hardware selection and software systems, and interaction modalities used in constructing the Silo. To address missing visual information due to the absence of a ceiling and floor, we have designed a method that utilizes conformal mapping and optimal mass transport to reproject the entire 360 volumetric FoR of the virtual scene to the available display real estate. We showcase several applications demonstrating the utility of the Silo and report the findings of our user studies that highlight the effectiveness of the Silo layout compared to curved mono and flat powerwall display facilities. Our user evaluations and studies have shown that the Silo supports natural exploration and enhanced visualization due to its capability to render surround ultra-high-resolution stereoscopic views.

关键词： visualization Three-dimensional displays Stereo image processing Data visualization Transportation Virtual reality Software systems Liquid crystal displays Floors Next generation networking