Multi-Scale Feature Fusion

Follow this topic Ask about this topic

Multi-scale feature fusion studies how to combine coarse contextual information with fine spatial, temporal, or spectral details. Common approaches include feature pyramids and skip connections, attention or gated fusion, cross-scale alignment, and adaptive routing across CNN, Transformer, and state-space models. These methods are used in settings such as small-object detection, medical and remote-sensing segmentation, time-series forecasting, and multimodal perception. A central challenge is preserving useful detail while aligning scales and avoiding redundant computation, with results often depending on the task, data, and model size.

Latest papers

187 papers14 papers in the last four weeks, down 30% on the four weeks before. 0.1% of all new papers.

DMSC: Dynamic Multi-Scale Coordination Framework for Time Series Forecasting
Aug 3, 2025Haonan Yang, Jianchao Tang, Zhuo Li +1Time Series Forecasting Multi-Scale Feature Fusion
National University of Defense Technology, Changsha, Hunan, China
MGDFIS: Multi-scale Global-detail Feature Integration Strategy for Small Object Detection
Jun 15, 2025Yuxiang Wang, Xuecheng Bai, Chuanzhi Xu +2Remote Sensing Small Object Detection Small Object Detection
School of Computer Science, The University of Sydney, Sydney, Australia · 2Aviation Traffic Control Technology (SHENZHEN) Co., Ltd., Shenzhen, China
Intra- and Cross-Scale Feature Interaction Network for Referring Remote Sensing Image Segmentation
Jun 4, 2025Zhigang Yang, Huiguang Yao, Linmao Tian +3Referring Image Segmentation Remote Sensing Image Segmentation
School of Artificial Intelligence, Optics and Electronics (iOPEN), Northwestern Polytechnical University, Xi’an 710072, P.R. China · School of Computer Science, Northwestern Polytechnical University, Xi’an 710072, P.R. China · department of Software, Northwest Institute of Nuclear Technology, Xi’an 710072, P.R. China
SMDDFNet: State-space Modeling and Dynamic Dual Fusion Network for Traffic Sign Detection
May 2, 2025TianYi Yu, DaJian Zhong, Lilin WangTraffic Sign Detection Small Object Detection
Collage of Infomation Engineering, Shanghai Maritime University, Shanghai, 201306, China
MSA-UNet3+: Multi-Scale Attention UNet3+ with New Supervised Prototypical Contrastive Loss for Coronary DSA Image Segmentation
Apr 7, 2025Rayan Merghani Ahmed, Adnan Iltaf, Mohamed Elmanna +5Contrastive Learning Medical Imaging
Shenzhen Institutes of Advanced Technology (SIAT), Chinese Academy of Sciences (CAS), Shenzhen, 518055, China · University of Chinese Academy of Sciences (UCAS), Beijing, 101408, China · Department of Biomedical Engineering and Systems, Faculty of Engineering, Cairo University, Cairo, 12613, Egypt +2
CrossFusion: A Multi-Scale Cross-Attention Convolutional Fusion Model for Cancer Survival Prediction
Mar 3, 2025Rustin Soraki, Huayu Wang, Sitong Liu +2Computational Pathology Cancer Survival Prediction
University of Washington, Seattle, WA · University of California, Los Angeles, CA
Cascaded Multi-Scale Attention for Enhanced Multi-Scale Feature Extraction and Interaction with Low-Resolution Images
Dec 3, 2024Xiangyong Lu, Masanori Suganuma, Takayuki OkataniHybrid CNN-Transformer Architectures Object Pose Estimation
Graduate School of Information Sciences, Tohoku University. · RIKEN Center for AIP

Aug 3, 2025J/K move · Enter open · S save

DMSC: Dynamic Multi-Scale Coordination Framework for Time Series Forecasting

Haonan Yang, Jianchao Tang, Zhuo Li, Long Lan

National University of Defense Technology, Changsha, Hunan, China

Time Series Forecasting Multi-Scale Feature Fusion Time Series Decomposition Multiscale Time-Series Modeling

Open paper AskPDF arXiv

Time Series Forecasting (TSF) faces persistent challenges in modeling intricate temporal dependencies across different scales. Despite recent advances leveraging different decomposition operations and novel architectures based on CNN, MLP or Transformer, existing methods still struggle with static decomposition strategies, fragmented dependency modeling, and inflexible fusion mechanisms, limiting their ability to model intricate temporal dependencies. To explicitly solve the mentioned three problems respectively, we propose a novel Dynamic Multi-Scale Coordination Framework (DMSC) with Multi-Scale Patch Decomposition block (EMPD), Triad Interaction Block (TIB) and Adaptive Scale Routing MoE block (ASR-MoE). Specifically, EMPD is designed as a built-in component to dynamically segment sequences into hierarchical patches with exponentially scaled granularities, eliminating predefined scale constraints through input-adaptive patch adjustment. TIB then jointly models intra-patch, inter-patch, and cross-variable dependencies within each layer's decomposed representations. EMPD and TIB are jointly integrated into layers forming a multi-layer progressive cascade architecture, where coarse-grained representations from earlier layers adaptively guide fine-grained feature extraction in subsequent layers via gated pathways. And ASR-MoE dynamically fuses multi-scale predictions by leveraging specialized global and local experts with temporal-aware weighting. Comprehensive experiments on thirteen real-world benchmarks demonstrate that DMSC consistently maintains state-of-the-art (SOTA) performance and superior computational efficiency for TSF tasks. Code is available at https://github.com/1327679995/DMSC.