arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~

期刊&会议

Conference on Computer Vision and Pattern Recognition · 会议 · Computer Vision

共收录 11901
2609.00610 2026-09-11 cs.CV 版本更新

Streaming4D: Accelerate 4D World Models via Block-wise Video Generation and Incremental Reconstruction

Streaming4D:通过分块视频生成与增量重建加速4D世界模型

Xiaoyan Liu, Jiaxin Liu, Kangrui Li, Sifan Zhou

机构 * The Chinese University of Hong Kong(香港中文大学) The Hong Kong Polytechnic University(香港理工大学) The University of New South Wales(新南威尔士大学) Southeast University(东南大学)

AI总结 Streaming4D提出分块视频生成与增量重建结合的同步流水线,在RTX 4090上实现1.24倍运行时加速,同时保持4D几何与多视图一致性,解决传统4D生成延迟高的问题。

Comments Accepted by CVPR 2026 4DV Workshop

详情

展开后加载摘要…

URL PDF HTML 收藏
2602.20328 2026-09-10 cs.CV eess.IV math.OC

GSNR: Graph Smooth Null-Space Representation for Inverse Problems

GSNR:图平滑空域表示用于逆问题

Romario Gualdrón-Hurtado, Roman Jacome, Rafael S. Suarez, Henry Arguello

机构 * Universidad Industrial de Santander(圣安德烈大学)

AI总结 本文提出GSNR方法,通过在不可见成分中施加结构,解决逆问题中空域信息约束不足的问题,提升重建精度。

Comments Accepted to The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2026 (CVPR 2026)

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026, pp. 12469-12479

详情

展开后加载摘要…

URL PDF HTML 收藏
2604.27715 2026-09-09 cs.CV

Improving Calibration in Test-Time Prompt Tuning for Vision-Language Models via Data-Free Flatness-Aware Prompt Pretraining

通过数据无关的平坦性感知提示预训练提升视觉语言模型测试时提示调优的校准

Hyeonseo Jang, Jaebyeong Jeon, Joong-Won Hwang, Kibok Lee

机构 * Yonsei University(延世大学) ETRI(韩国电子技术研究院)

AI总结 本文提出FPP框架,通过在提示初始化时选择平坦区域,提升测试时提示调优的校准和性能,无需额外标注数据和计算成本。

Comments CVPR 2026

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026, pp. 24300-24309

详情

展开后加载摘要…

URL PDF HTML 收藏
2604.18075 2026-09-09 cs.CV

Enhancing Continual Learning of Vision-Language Models via Dynamic Prefix Weighting

通过动态前缀加权增强视觉-语言模型的持续学习

Hyeonseo Jang, Hyuk Kwon, Kibok Lee

机构 * Yonsei University(延世大学)

AI总结 本文提出动态前缀加权框架,通过动态调整前缀权重和适配器机制,提升视觉-语言模型在领域-类别增量学习中的性能。

Comments CVPR 2026; revised text and figures for improved readability

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026, pp. 18042-18052

详情

展开后加载摘要…

URL PDF HTML 收藏
2604.15311 2026-09-09 cs.CV 版本更新

LeapAlign: Post-Training Flow Matching Models at Any Generation Step by Building Two-Step Trajectories

LeapAlign: 通过构建两步轨迹实现任意生成步骤的训练后流匹配模型对齐

Zhanhao Liang, Tao Yang, Jie Wu, Chengjian Feng, Liang Zheng

机构 * The Australian National University(澳大利亚国立大学)

AI总结 本文提出LeapAlign方法,通过缩短生成轨迹为两步,减少计算成本并实现早期生成步骤的梯度传播,提升模型更新效率与稳定性,优于现有方法。

Comments Accepted by CVPR 2026. Project page: https://rockeycoss.github.io/leapalign/

详情

展开后加载摘要…

URL PDF HTML 收藏
2604.12403 2026-09-09 cs.CV 版本更新

Dual-Modality Anchor-Guided Filtering for Test-time Prompt Tuning

双模锚引导过滤用于测试时提示微调

Jungwon Choi, Eunwoo Kim

机构 * School of Computer Science and Engineering, Chung-Ang University(Chung-Ang大学计算机科学与工程学院)

AI总结 本文提出双模锚引导框架,通过语义证据指导视图选择,结合文本和图像锚点进行过滤和集成,提升测试时提示微调的鲁棒性。

Comments Accepted by CVPR 2026 findings

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.03619 2026-09-09 cs.CV 版本更新

LAMP: Language-Assisted Motion Planning for Controllable Video Generation

LAMP:语言辅助运动规划用于可控视频生成

Muhammed Burak Kizil, Enes Sanli, Niloy J. Mitra, Erkut Erdem, Aykut Erdem, Duygu Ceylan

机构 * Koç University(科奇大学) University College London(伦敦大学学院) Hacettepe University(哈斯特帕大学) Adobe Research(Adobe研究院)

AI总结 LAMP通过大语言模型生成3D轨迹,实现基于自然语言的视频生成,提升运动可控性和用户意图对齐。

Comments CVPR 2026. Project Page: https://cyberiada.github.io/LAMP/

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.18493 2026-09-09 eess.IV cs.AI cs.CV 版本更新

SAGE: Shape-Adapting Gated Experts for Adaptive Histopathology Image Segmentation

SAGE:适应性组织病理图像分割的形状自适应门控专家

Gia Huy Thai, Hoang-Nguyen Vu, Anh-Minh Phan, Quang-Thinh Ly, Tram Dinh, Thi-Ngoc-Truc Nguyen, Nhat Ho

机构 * University of Science, VNU-HCM(越南国家大学科学学院) Trivita AI University of Technology, VNU-HCM(越南国家大学技术学院) Michigan State University, USA(美国密歇根州立大学) The University of Texas at Austin(德克萨斯大学奥斯汀分校)

AI总结 SAGE通过动态专家路由框架提升异构视觉网络中细胞形态变化的适应性,实现高精度分割与稳健泛化。

Comments Accepted to CVPR 2026 (Findings Track). Project Page: https://oxyzgiahuy.github.io/sage/

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.12013 2026-09-09 cs.CV

Pano360: Perspective to Panoramic Vision with Geometric Consistency

Pano360: 透视到全景视觉的几何一致性

Zhengdong Zhu, Weiyi Xue, Zuyuan Yang, Wenlve Zhou, Zhiheng Zhou

机构 * South China University of Technology(华南理工大学) Tongji University(同济大学) Guangdong University of Technology(广东工业大学)

AI总结 Pano360通过引入3D摄影测量空间和基于变换器的架构,实现了更准确的全景拼接,提升了对齐精度和视觉质量。

Comments Accepted by CVPR2026

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026, pp. 7600-7609

详情

展开后加载摘要…

URL PDF HTML 收藏
2406.04314 2026-09-09 cs.CV 版本更新

Aesthetic Post-Training Diffusion Models from Generic Preferences with Step-by-step Preference Optimization

基于通用偏好的美学后训练扩散模型与逐步偏好优化

Zhanhao Liang, Yuhui Yuan, Shuyang Gu, Bohan Chen, Tiankai Hang, Mingxi Cheng, Ji Li, Liang Zheng

机构 * The Australian National University(澳大利亚国立大学) University of Liverpool(利物浦大学) Southeast University(东南大学) Microsoft(微软公司) Microsoft Research Asia(微软亚洲研究院)

AI总结 本文提出逐步偏好优化(SPO),利用通用偏好数据,通过步骤感知偏好模型和候选采样,在去噪各步骤细粒度监督扩散模型,显著提升美学质量且不牺牲对齐,收敛更快。

Comments CVPR 2025. Project Page: https://rockeycoss.github.io/spo.github.io/

详情

展开后加载摘要…

URL PDF HTML 收藏
2410.13924 2026-09-09 cs.CV cs.AI

ARKit LabelMaker: A New Scale for Indoor 3D Scene Understanding

Guangda Ji, Silvan Weder, Francis Engelmann, Marc Pollefeys, Hermann Blum

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025, pp. 4398-4407

详情

展开后加载摘要…

URL PDF HTML 收藏
2503.04666 2026-09-09 cs.CV 版本更新

What Are You Doing? A Closer Look at Controllable Human Video Generation

你在做什么?可控人类视频生成的深入探究

Emanuele Bugliarello, Anurag Arnab, Roni Paiss, Christy Koh, Pieter-Jan Kindermans, Cordelia Schmid

AI总结 针对现有数据集缺乏多样性,提出WYD基准,含1544个视频和细粒度标注,评估可控人类视频生成的9个方面,并改进评估指标,深入分析开源模型,发布数据与代码。

Comments CVPR 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2104.00239 2026-09-09 cs.CV cs.MM cs.SD eess.AS 版本更新

Positive Sample Propagation along the Audio-Visual Event Line

正样本沿音视频事件线的传播

Jinxing Zhou, Liang Zheng, Yiran Zhong, Shijie Hao, Meng Wang

机构 * Hefei University of Technology(合肥工业大学) Intelligent Interconnected Systems Laboratory of Anhui Province(安徽省智能互联系统实验室) Australian National University(澳大利亚国立大学)

AI总结 本文提出正样本传播模块,通过全对相似度图与相似度损失,在音视频事件定位中有效利用正样本对,并在AVE数据集上取得最先进结果。

Comments Accepted to CVPR 2021. Code is available at https://github.com/jasongief/PSP_CVPR_2021

详情

展开后加载摘要…

URL PDF HTML 收藏
2609.04550 2026-09-07 cs.CV 新提交

VISTA: Dense Multi-Label Classroom Coding with Vision-Language Models

VISTA:基于视觉语言模型的密集多标签课堂编码

Andrew Franck, Brendan Ng, Ben Fitzgerald, Zane Derrod, Chris Cianci, Chris Craney

机构 * Occidental College(西方学院)

AI总结 本研究将COPUS构建为多模态基准,提出基于MiniCPM-V-4.5和MLP的VISTA基线,在化学讲座测试中受限宏准确率达80.1%,同时发布相关工具与代码。

Comments DataMFM Workshop @ Computer Vision & Pattern Recognition (CVPR) 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2604.12762 2026-09-07 cs.CV cs.AI cs.MA 版本更新

ARGOS: Who, Where, and When in Agentic Multi-Camera Person Search

ARGOS:智能多摄像机人像搜索中的谁、哪里和何时

Myungchul Kim, Kwanyong Park, Junmo Kim, In So Kweon

机构 * KAIST(韩国科学技术院) University of Seoul(首尔大学) KIST(韩国科学技术院)

AI总结 ARGOS首次将多摄像机人像搜索转化为交互推理问题,通过智能体在信息不对称下规划、提问和消除候选,包含14个真实场景的2691个任务,实验表明该基准远未解决。

Comments Accepted to ECCV 2026 & CVPR 2026 Workshop on Multimodal Spatial Intelligence (MUSI)

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.26425 2026-09-07 cs.CV cs.AI

CPUBone: Efficient Vision Backbone Design for Devices with Low Parallelization Capabilities

CPUBone:为低并行化能力设备设计的高效视觉骨干网络

Moritz Nottebaum, Matteo Dunnhofer, Christian Micheloni

机构 * University of Udine, Italy(意大利乌迪内大学) York University, Canada(加拿大约克大学)

AI总结 本文提出CPUBone,一种针对CPU推理优化的视觉骨干网络,通过改进卷积操作降低计算成本,实现高效的Speed-Accuracy Trade-offs。

Comments Accepted at CVPR Findings 2026

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Findings, 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2505.18686 2026-09-07 cs.CV 版本更新

WeakMCN: Multi-task Collaborative Network for Weakly Supervised Referring Expression Comprehension and Segmentation

WeakMCN:用于弱监督指代表达理解与分割的多任务协作网络

Silin Cheng, Yang Liu, Xinwei He, Sebastien Ourselin, Lei Tan, Gen Luo

AI总结 该研究针对弱监督指代表达理解与分割任务,提出多任务协作网络WeakMCN,通过双分支架构及两项创新设计提升性能,在多个基准上优于单任务方法且泛化能力强。

Comments Accepted by CVPR 2025

详情

展开后加载摘要…

URL PDF HTML 收藏
2609.03302 2026-09-04 cs.CV q-bio.NC 新提交

Tensor-based Brain Surface Modeling and Analysis

基于张量的脑表面建模与分析

Moo K. Chung, Keith J. Worsley, Steve Robbins, Alan C. Evans

机构 * University of Wisconsin-Madison(威斯康星大学麦迪逊分校) Montreal Neurological Institute, McGill University(麦吉尔大学蒙特利尔神经学研究所)

AI总结 该研究提出统一计算框架,结合表面建模、平滑与统计分析,通过张量形态测量学定位儿童脑皮层灰质变化区域,实现脑表面形状差异检测。

Journal ref The proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2003 Vol I 467-473

详情

展开后加载摘要…

URL PDF HTML 收藏
2609.03102 2026-09-04 cs.CV 新提交

WireSeg-32K: A Physics-Grounded Synthetic Dataset for Wire Instance Segmentation

WireSeg-32K:用于导线实例分割的基于物理的合成数据集

Zilin Dai, Lehong Wang, Yi Yang, Xiang Fei

机构 * Harvard University(哈佛大学) Carnegie Mellon University(卡内基梅隆大学)

AI总结 针对导线等可变形线性物体分割难、真实场景标注成本高的问题,提出含32000张图像的WireSeg-32K合成数据集,开发DeformX协同仿真流水线生成数据,经LoRA微调SAM3后真实场景mAP@75提升10.2%,验证了该合成数据的迁移价值。

Comments Synthetic Data for Computer Vision @ CVPR 2026 Workshop

详情

展开后加载摘要…

URL PDF HTML 收藏
2609.01816 2026-09-03 cs.CV 新提交

Video2Reaction: Training Foundation Video Models to Predict Audience Reaction

Video2Reaction:训练基础视频模型以预测观众反应

Sidong Zhang, Trang Nguyen, Shiv Shankar, Gauri Jagatap, Deepak Chandran, Andrea Fanelli, Madalina Fiterau

机构 * UMass Amherst(马萨诸塞大学阿默斯特分校) Dolby Laboratories(杜比实验室)

AI总结 本研究构建多模态数据集Video2Reaction,对采用LoRA微调的VLMs进行基准测试,发现其可有效学习并迁移至VCE数据集,仅用1%VCE训练数据适配的LLaVA-NeXT-Video-7B可达到与全数据训练相当的性能。

Comments Presented in the Workshop on Emerging Directions in Data for Multimodal Foundation Models at CVPR 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2507.14811 2026-09-03 cs.CV cs.AI

SegQuant: A Semantics-Aware and Generalizable Quantization Framework for Diffusion Models

SegQuant: 一种语义感知且通用的扩散模型量化框架

Jiaji Zhang, Ruichao Sun, Hailiang Zhao, Jiaju Wu, Peng Chen, Hao Li, Yuying Liu, Kingsum Chow, Gang Xiong, Shuiguang Deng

AI总结 SegQuant提出一种统一的量化框架,通过自适应结合互补技术提升跨模型通用性,采用结构感知图量化策略和双尺度量化方案,有效保持视觉保真度,适用于Transformer扩散模型及其他模型。

Comments 22 pages, 15 figures, to be published in The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2026 (CVPR 2026), code is available at https://github.com/OptiSys-ZJU/segquant

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026, pp. 43717-43726

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.02329 2026-09-03 cs.CV

HAMMER: Harnessing MLLM via Cross-Modal Integration for Intention-Driven 3D Affordance Grounding

HAMMER: 通过跨模态整合利用大语言模型进行意图驱动的3D affordance grounding

Lei Yao, Yong Chen, Yuejiao Su, Yi Wang, Moyun Liu, Lap-Pui Chau

机构 * The Hong Kong Polytechnic University(香港理工大学) Huazhong University of Science and Technology(华中科技大学)

AI总结 HAMMER通过跨模态整合多模态大语言模型,实现意图驱动的3D affordance grounding,提升3D表示的准确性和鲁棒性。

Comments Accepted by CVPR 2026. Project Page: https://rayyoh.github.io/Hammer

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026, pp. 23805-23815

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.09681 2026-09-03 cs.LG cs.AI 版本更新

SEBA: Sample-Efficient Black-Box Attacks on Visual Reinforcement Learning

SEBA:针对视觉强化学习的样本高效黑盒攻击

Tairan Huang, Yulin Jin, Junxu Liu, Qingqing Ye, Haibo Hu

机构 * The Hong Kong Polytechnic University(香港理工大学)

AI总结 SEBA是针对视觉强化学习的样本高效黑盒攻击框架,通过影子Q模型、生成对抗网络与世界模型的结合,在MuJoCo和Atari基准上实现了高效且低感知的攻击,大幅减少环境交互。

Comments Accepted to CVPR 2026

Journal ref Huang T, Jin Y, Liu J, et al. SEBA: Sample-Efficient Black-Box Attacks on Visual Reinforcement Learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2026: 27861-27871

详情

展开后加载摘要…

URL PDF HTML 收藏
2602.24136 2026-09-02 cs.CV

Prune Wisely, Reconstruct Sharply: Compact 3D Gaussian Splatting via Adaptive Pruning and Difference-of-Gaussian Primitives

明智剪枝,精确重建:通过自适应剪枝和高斯差分原语实现紧凑的3D高斯溅射

Haoran Wang, Guoxi Huang, Fan Zhang, David Bull, Nantheera Anantrasirichai

机构 * School of Computer Science, University of Bristol(计算机科学学院,布里斯托大学)

AI总结 本文提出了一种自适应剪枝策略和高斯差分原语,通过减少高斯数量提升3D高斯溅射的紧凑性与渲染质量。

Comments CVPR2026

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 11716-11725, 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2503.03637 2026-09-02 cs.CV eess.IV

L2RDaS: Synthesizing 4D Radar Tensors for Model Generalization via Dataset Expansion

Woo-Jin Jung, Dong-Hee Paek, Seung-Hyun Kong

机构 * CCS Graduate School of Mobility, Korea Advanced Institute of Science and Technology(移动研究学院,韩国科学技术院)

Comments 9 pages, 3 figures, Arxiv preprint

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Findings, 2026, pp. 889-899

详情

展开后加载摘要…

URL PDF HTML 收藏
2608.30653 2026-09-01 cs.CV cs.AI cs.LG 新提交

Fine-Grained Multi Image Object Hallucination Benchmark

细粒度多图像物体幻觉基准测试

Joonki Min, Chaeyun Kim, Hyungwook Choi, Yejin Kim, Kihyun Kim, Yohan Jo, Joonseok Lee

机构 * Seoul National University(首尔大学) AIM Intelligence(AIM智能公司)

AI总结 本研究针对多模态大语言模型的物体幻觉问题,推出细粒度多图像物体幻觉基准测试MIOH,评估29个模型后发现顶尖模型仍存在明显失败模式,为开发可靠多模态AI提供关键评估工具。

Comments Accepted at CVPR 2026

Journal ref Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026, pp. 18295-18305

详情

展开后加载摘要…

URL PDF HTML 收藏
2608.29759 2026-09-01 cs.CV cs.AI 新提交

SynCrash: A Multi-Stage Pipeline for Zero-Shot Accident Detection and Localization in Traffic Surveillance Video

SynCrash:面向交通监控视频零样本事故检测与定位的多阶段流水线

Arkya Jyoti Bagchi, Ritul Jangir, Varun Raskar

机构 * Indian Institute of Technology Jodhpur(印度焦特布尔印度理工学院)

AI总结 提出SynCrash多阶段流水线,针对CVPR2026挑战赛ACCIDENT任务,在无真实标注数据下,结合VideoMAEv2、YOLO及物理启发式算法,实现交通监控视频零样本事故检测、定位与碰撞分类。

Comments Accepted at the CVPR 2026 AUTOPILOT Workshop (non-archival)

详情

展开后加载摘要…

URL PDF HTML 收藏
2608.29252 2026-09-01 cs.AI 新提交

Dynamic Important Example Mining for Reinforcement Finetuning

用于强化微调的动态重要示例挖掘

Haoru Tan, Sitong Wu, Yanfeng Chen, Shizhen Zhao, Yang-Tian Sun, Tianjia Liu, Chirui Chang, Shaofeng Zhang, Samm Sun, Xiuzhe Wu, Ruobing Xie, Xiaojuan Qi

机构 * HKU(香港大学) Tencent(腾讯) CUHK(香港中文大学) Stanford(斯坦福大学)

AI总结 该研究针对强化微调中样本价值固定假设的缺陷,提出DIEM框架,整合梯度对齐重要性估计器与约束批量重加权方案,在多个推理基准中优于相关基线。

Journal ref CVPR-2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2606.07288 2026-09-01 cs.CV cs.GR 版本更新

ExMesh: Explicit Mesh Reconstruction with Topology Adaptation

ExMesh: 具有拓扑自适应的显式网格重建

Chuanjin Fan, Lifan Wu, Wenjie Chang, Hanzhi Chang, Wenfei Yang, Tianzhu Zhang

机构 * University of Science and Technology of China(中国科学技术大学) National Key Laboratory of Deep Space Exploration, Deep Space Exploration Laboratory(国家空间科学探测重点实验室,深空探测实验室)

AI总结 提出ExMesh框架,通过可微优化与离散拓扑更新直接优化显式网格,引入自适应顶点分裂合并和实时UV维护,实现从粗到细的优化,兼顾精度、效率和网格简洁性。

Comments Accepted at the IEEE/CVF Conference on Computer Vision and Pattern Recognition 2026 (CVPR 2026); Project page: https://fan-treasure.github.io/ExMesh_page.github.io/

详情

展开后加载摘要…

URL PDF HTML 收藏
2605.22455 2026-09-01 cs.CV cs.AI cs.LG physics.optics 版本更新

Making the Discrete Continuous: Synthetic RAW Augmentations for Fine-Grained Evaluation of Person Detection Performance in Low Light

使离散的成为连续的:合成RAW增强用于细粒度评估人检测性能在低光环境

Valeria Pais, Malena Mendilaharzu, Daniele Faccio, Luis Oala, Christoph Clausen, Bruno Sanguinetti

机构 * University of Glasgow(格拉斯哥大学) Dotphoton

AI总结 本文提出了一种合成RAW增强方法,用于在低光条件下更准确地评估人检测模型的性能,通过生成与相机传感器噪声模型匹配的低光样本,以改善基准测试的数据覆盖。

Comments Accepted non-archival paper at the CVPR 2026 AUTOPILOT Workshop (Autonomous Understanding Through Open-world Perception and Integrated Language Models for On-road Tasks)

详情

展开后加载摘要…

URL PDF HTML 收藏