NSTL国家科技图书文献中心

1. Hybrid Scalable Video Coding with Neural Compression and Enhancement for Streaming Media NSTL国家科技图书文献中心

Yuyao Ye | Jiayu Yang ... - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 74～86 - 共13页

摘要：Streaming media is important in modern | information consumption industry. However the limited and | varied computing resources and bandwidth of client | devices pose challenges for video coding. To make a | balance between speed and coding efficiency and provide

关键词： Video coding | Scalable codec | Video enhancement

2. Lightweight Dual Grouped Large-Kernel Convolutions for Salient Object Detection Network NSTL国家科技图书文献中心

Jiajie Liu | Zhibin Zhang - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 240～253 - 共14页

摘要：Most existing Salient Object Detection (SOD | ) methods focus on achieving better performance, often | resulting in models with a large number of parameters | . However, there is limited research on lightweight models | in this field. To address this gap, our goal is to

关键词： Segmentation | Matting | Lightweight network

3. KuzushijiDiffuser: Japanese Kuzushiji Font Generation with FontDiffuser NSTL国家科技图书文献中心

Honghui Yuan | Keiji Yanai - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 212～225 - 共14页

摘要：Kuzushiji characters were used in Japan | hundreds of years ago, and many valuable ancient | documents are written in Kuzushiji. Research into | generating Kuzushiji characters increases the training data | for recognizing these characters and enhances

关键词： Kuzushiji characters | Font generation | FontDiffuser

4. MineTinyNet-YOLO: An Efficient Small Object Detection Method for Complex Underground Coal Mine Scenarios NSTL国家科技图书文献中心

Yaling Hao | Wei Wu - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 364～378 - 共15页

摘要：The YOLO series of algorithms has become the | primary method for real-time object detection. Many | studies have enhanced the benchmark performance by | adjusting model structures, updating training methods, and | optimizing hyperparameters. However, in underground coal

关键词： Complex backgrounds | Small target detection | MineTinyNet-YOLO | DBACL | AIF-Net | PUO

5. MC-YOLO: Multi-scale Transmission Line Defect Target Recognition Network NSTL国家科技图书文献中心

Jingdong Wang | Xu Ding ... - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 324～337 - 共14页

摘要：Aiming at addressing the challenges posed by | complex backgrounds in the inspection process of | transmission lines, as well as the poor detection effect | resulting from a high number of targets with varying | scales and the tendency to overlook or incorrectly

关键词： Transmission line defect detection | Multi-scale fusion | PKI module | MSF

6. Lightweight Motion-Aware Video Super-Resolution for Compressed Videos NSTL国家科技图书文献中心

Ilhwan Kwon | Jun Li ... - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 254～267 - 共14页

摘要：This study introduces a lightweight and | efficient Video Super-Resolution (VSR) model that | leverages codec prior information, focusing on motion | . Unlike existing VSR models that aim for superior | performance through complex structures, the proposed model

关键词： Video super-Resolution | Video compression | Lightweight

7. Hyper-NeuS: Hypernetworks for Neural SDF Implicit Surface Reconstruction by Volume Rendering NSTL国家科技图书文献中心

Jingkun Li | Na Qi ... - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 87～100 - 共14页

摘要：Neural Radiance Field (NeRF) has gained | widespread attention in the research of novel view | synthesis and 3D shape reconstruction due to their | powerful rendering capabilities in recent years. Although | volume density of NeRF can reconstruct surface geometry

关键词： Scene reconstruction | Neural radiance field | Hypernetworks | Signed distance field

8. MKSNet: Advanced Small Object Detection in Remote Sensing Imagery with Multi-Kernel and Dual Attention Mechanisms NSTL国家科技图书文献中心

Jiahao Zhang | Xiao Zhao ... - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 394～407 - 共14页

摘要：Deep convolutional neural networks (DCNNs | ) have substantially advanced object detection | capabilities, particularly in remote sensing imagery. However | , challenges persist, especially in detecting small objects | where the high resolution of these images and the

关键词： Remote sensing images | Small object detection | Multi-Kernel selection | Spatial attention | Channel attention

9. Integrating S1&S2 Framework for Enhanced Semantic Match in Person Re-identification NSTL国家科技图书文献中心

Xiukang Yang | Jingguo Ge ... - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 155～168 - 共14页

摘要：Person re-identification (ReID) is a | challenging task in computer vision, aimed at recognizing | and tracking individuals across different scenes | . The traditional approach involves feature extraction | , preliminary sorting, leading to potential mismatches with

关键词： ReID | S1&S2 | Multimodal

10. HCV: Lightweight Hybrid CNN-Vision Transformer for Visual Object Tracking NSTL国家科技图书文献中心

Liang-Chia Chen | Wei-Ta Chu - 《MultiMedia Modeling,Part II》 - International Conference on MultiMedia Modeling - 2025, - 45～59 - 共15页

摘要：Visual object tracking is one of the most | fundamental research in computer vision. Recent mainstream | trackers prioritize accuracy, leading to issues such as | prolonged computation time and substantial computational | resources to achieve significant performance. To address

摘要：Streaming media is important in modern | information consumption industry. However the limited and | varied computing resources and bandwidth of client | devices pose challenges for video coding. To make a | balance between speed and coding efficiency and provide

关键词： Video coding | Scalable codec | Video enhancement

摘要：Most existing Salient Object Detection (SOD | ) methods focus on achieving better performance, often | resulting in models with a large number of parameters | . However, there is limited research on lightweight models | in this field. To address this gap, our goal is to

关键词： Segmentation | Matting | Lightweight network

摘要：Kuzushiji characters were used in Japan | hundreds of years ago, and many valuable ancient | documents are written in Kuzushiji. Research into | generating Kuzushiji characters increases the training data | for recognizing these characters and enhances

关键词： Kuzushiji characters | Font generation | FontDiffuser

摘要：The YOLO series of algorithms has become the | primary method for real-time object detection. Many | studies have enhanced the benchmark performance by | adjusting model structures, updating training methods, and | optimizing hyperparameters. However, in underground coal

关键词： Complex backgrounds | Small target detection | MineTinyNet-YOLO | DBACL | AIF-Net | PUO

摘要：Aiming at addressing the challenges posed by | complex backgrounds in the inspection process of | transmission lines, as well as the poor detection effect | resulting from a high number of targets with varying | scales and the tendency to overlook or incorrectly

关键词： Transmission line defect detection | Multi-scale fusion | PKI module | MSF

摘要：This study introduces a lightweight and | efficient Video Super-Resolution (VSR) model that | leverages codec prior information, focusing on motion | . Unlike existing VSR models that aim for superior | performance through complex structures, the proposed model

关键词： Video super-Resolution | Video compression | Lightweight

摘要：Neural Radiance Field (NeRF) has gained | widespread attention in the research of novel view | synthesis and 3D shape reconstruction due to their | powerful rendering capabilities in recent years. Although | volume density of NeRF can reconstruct surface geometry

关键词： Scene reconstruction | Neural radiance field | Hypernetworks | Signed distance field

摘要：Deep convolutional neural networks (DCNNs | ) have substantially advanced object detection | capabilities, particularly in remote sensing imagery. However | , challenges persist, especially in detecting small objects | where the high resolution of these images and the

关键词： Remote sensing images | Small object detection | Multi-Kernel selection | Spatial attention | Channel attention

摘要：Person re-identification (ReID) is a | challenging task in computer vision, aimed at recognizing | and tracking individuals across different scenes | . The traditional approach involves feature extraction | , preliminary sorting, leading to potential mismatches with

关键词： ReID | S1&S2 | Multimodal

摘要：Visual object tracking is one of the most | fundamental research in computer vision. Recent mainstream | trackers prioritize accuracy, leading to issues such as | prolonged computation time and substantial computational | resources to achieve significant performance. To address

关键词： Visual object tracking | Lightweight tracking | Hierarchical features | Vision transformer

4008-161-200 800-990-8900

国家科技图书文献中心