IEEE Transactions on Circuits and Systems for Video Technology最新文献_第9页

Video Corpus Moment Retrieval via Decoupled Multimodal Modeling and Unified Localization 基于解耦多模态建模和统一定位的视频语料库矩检索

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-22 DOI: 10.1109/TCSVT.2026.3705906

Yong Yang;Meng Liu;Xuemeng Song;Na Zheng;Ke Lu;Weili Guan

{"title":"Video Corpus Moment Retrieval via Decoupled Multimodal Modeling and Unified Localization","authors":"Yong Yang;Meng Liu;Xuemeng Song;Na Zheng;Ke Lu;Weili Guan","doi":"10.1109/TCSVT.2026.3705906","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3705906","url":null,"abstract":"Video Corpus Moment Retrieval (VCMR) aims to retrieve the target temporal moment from a large-scale corpus of untrimmed videos given a natural-language query. A central challenge of this task lies in effectively modeling heterogeneous video modalities, such as visual content and subtitles, together with the query. Existing methods typically fuse multiple video modalities before aligning the fused representation with the query. Although effective under the full-modality setting, such tightly coupled modeling may obscure modality-specific discriminative cues and reduce robustness when modalities are noisy, imbalanced, or unavailable. To address this issue, we propose DeMUL, a new VCMR framework with <italic>Decoupled Multimodal Modeling and <italic>Unified Localization. Instead of directly fusing different video modalities, DeMUL first aligns the query with each modality separately and then integrates the resulting query-aligned modality-specific representations for temporal localization. This decoupled design preserves modality-specific information while still exploiting complementary multimodal evidence. Building on this formulation, we further introduce a unified localization module that jointly optimizes unimodal and multimodal branches in a shared prediction space, enabling consistent localization behavior across branches without requiring auxiliary distillation or additional supervision. Extensive experiments on TVR and DiDeMo demonstrate that DeMUL achieves state-of-the-art performance under the standard full-modality setting, improving the performance from 18.95 to 21.37 on TVR (+2.42, 12.8%) and from 4.77 to 5.52 on DiDeMo (+0.75, 15.7%) compared with the previous state-of-the-art method QCLPL. In addition, DeMUL maintains strong performance under modality-missing and degraded-input conditions, demonstrating superior robustness in realistic multimodal retrieval scenarios. We have also released our code at <uri>https://github.com/Yongyangyyy/DeMUL</uri>.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15069-15081"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149096973","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Quality-Preserving Imperceptible Adversarial Attack on Skeleton-Based Human Action Recognition 基于骨骼的人体动作识别的保质不可察觉对抗攻击

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-12 DOI: 10.1109/TCSVT.2026.3703135

Ziyi Chang;Kanglei Zhou;Xiaohui Liang;Hubert P. H. Shum

{"title":"Quality-Preserving Imperceptible Adversarial Attack on Skeleton-Based Human Action Recognition","authors":"Ziyi Chang;Kanglei Zhou;Xiaohui Liang;Hubert P. H. Shum","doi":"10.1109/TCSVT.2026.3703135","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3703135","url":null,"abstract":"Adversarial attacks on skeletal human action recognition have received significant attention. However, existing methods typically introduce noise-like perturbations that degrade motion quality post-attack, and thereby are inherently perceptible with recent advancements in S-HAR systems. We discover that this degradation stems from the gap between empirical and true risks during the optimization process of previous adversarial attacks. To address this issue, we propose an attack where adversarial motions are obtained without compromising their motion quality. To minimize the risk gap and preserve motion quality, we propose a distribution-based adversarial attack method without introducing noise-like perturbations. To faithfully evaluate the motion quality, we propose a new metric that aligns with human perception on real-world naturalness. Experiments have been conducted on the state-of-the-art S-HAR methods across two datasets, demonstrating the superiority of our method in both the attack success rate and the post-attack motion quality through qualitative and quantitative analyses. The success of our quality-preserving attack application and distribution-based method raises serious concerns about the robustness of action recognizers, highlighting the need for further enhancements in this domain.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14520-14534"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097132","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

DeGMix: Efficient Multi-Task Dense Prediction With Deformable and Gating Mixer DeGMix：有效的多任务密集预测与可变形和浇注混合器

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-10 DOI: 10.1109/TCSVT.2026.3702343

Yangyang Xu;Yibo Yang;Bernard Ghanem;Lefei Zhang;Bo Du;Jun Zhu

{"title":"DeGMix: Efficient Multi-Task Dense Prediction With Deformable and Gating Mixer","authors":"Yangyang Xu;Yibo Yang;Bernard Ghanem;Lefei Zhang;Bo Du;Jun Zhu","doi":"10.1109/TCSVT.2026.3702343","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3702343","url":null,"abstract":"Convolution neural networks (CNNs) and Transformers offer complementary strengths for multi-task dense prediction: CNNs are effective at modeling local spatial patterns, while Transformers capture long-range dependencies. However, existing methods often rely primarily on one paradigm or do not fully exploit their complementarity for task-aware representation learning. In this work, we propose <bold>DeGMix (<bold>Deformable and <bold>Gating <bold>Mixer), a unified encoder–decoder framework for multi-task dense prediction that combines deformable convolution and attention-based task interaction. DeGMix consists of a deformable mixer encoder for task-aware spatial feature extraction and a task-aware gating Transformer decoder for cross-task interaction and task-specific feature selection. In this way, the proposed framework jointly models local deformable cues, global dependencies, and task-aware feature adaptation within a single architecture. Extensive experiments on three benchmark datasets, including NYUD-v2, PASCAL-Context, and Cityscapes, show that DeGMix consistently outperforms competitive CNN-based and Transformer-based multi-task methods across multiple dense prediction tasks. For example, with a Swin-L backbone, DeGMix achieves 57.55 mIoU on semantic segmentation on NYUD-v2, surpassing the previous best result by + 5.2 mIoU. In parameter-efficient fine-tuning on PASCAL-Context, DeGMix obtains a <inline-formula> <tex-math>$+ 1.96~Delta _{m}$ </tex-math></inline-formula> gain with 2.2M trainable parameters, approaching MTL-full fine-tuning (<inline-formula> <tex-math>$+ 2.23~Delta _{m}$ </tex-math></inline-formula>) while updating <inline-formula> <tex-math>$13.7times $ </tex-math></inline-formula> fewer parameters. Our code and models are publicly available at <uri>https://github.com/yangyangxu0/DeGMix</uri>","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15553-15567"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097090","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

OccM3D-SS: Bridging Sparse Supervision With Dense 3D Perception via Semi-Supervised Occupancy Learning OccM3D-SS：通过半监督占用学习架起稀疏监督与密集3D感知的桥梁

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-08-17 DOI: 10.1109/TCSVT.2026.3724359

Ke Liu;Xinzhu Ma;Ruxuan Hu;Boqian Liu;Zhiyong Wang;Zhihui Wang

{"title":"OccM3D-SS: Bridging Sparse Supervision With Dense 3D Perception via Semi-Supervised Occupancy Learning","authors":"Ke Liu;Xinzhu Ma;Ruxuan Hu;Boqian Liu;Zhiyong Wang;Zhihui Wang","doi":"10.1109/TCSVT.2026.3724359","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3724359","url":null,"abstract":"Monocular 3D object detection faces inherent challenges in reconstructing accurate 3D spatial structures from 2D images due to the lack of depth cues. Existing paradigms either struggle with limited 3D representation capability, <italic>e.g. depth estimation pipeline, or rely heavily on fine-grained supervision annotations, <italic>e.g. occupancy-guided pipeline. Moreover, occupancy labels derived from sparse LiDAR signals are often unreliable, introducing noise into geometric supervision. In this work, we propose <bold>OccM3D-SS, a semi-supervised occupancy-guided framework that integrates occupancy learning with unlabeled sequential data to enhance geometric reasoning. Our framework employs a dual-branch architecture: the supervised branch enforces semantic and geometric perception using annotated data, while the unsupervised branch leverages occupancy constraint to enhance geometric representation capability. Furthermore, to address the label noise issue, we introduce a temporal point-cloud densification strategy by tracking and aggregating foreground points from multi-frames, thereby improving the quality of occupancy annotations significantly. The experimental results on the KITTI-3D and Waymo Open datasets demonstrate that our framework outperforms the baseline by a large margin, achieving state-of-the-art performance while reducing annotation dependency.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14826-14837"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097109","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Transformer-Inspired Convolutional Network for Image Restoration 基于变压器的图像恢复卷积网络

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-07-06 DOI: 10.1109/TCSVT.2026.3710432

Yuning Cui;Mingyu Liu;Wenqi Ren;Boxin Shi;Alois Knoll

{"title":"Transformer-Inspired Convolutional Network for Image Restoration","authors":"Yuning Cui;Mingyu Liu;Wenqi Ren;Boxin Shi;Alois Knoll","doi":"10.1109/TCSVT.2026.3710432","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3710432","url":null,"abstract":"Image restoration is a critical task in computer vision, as unwanted degradations significantly reduce image quality and adversely affect the robustness of downstream vision systems. Recently, Transformer-based models have demonstrated remarkable performance in image restoration tasks. However, their quadratic computational complexity with respect to input size limits their practical application. In this paper, we develop a computationally efficient convolutional image restoration model inspired by the design philosophy of Transformer blocks. Specifically, we emulate and unify the functionalities of self-attention and feed-forward layers within a single convolution-based module to achieve adaptive and efficient spatial feature learning. Additionally, this module incorporates multiple receptive fields at attention, module, and scale levels, enabling effective management of degradations of varying sizes. Furthermore, to mitigate frequency discrepancies between degraded and sharp image pairs, we widen and deepen the structure of a frequency learning mechanism, which implicitly refines frequency spectra through a simple yet powerful frequency decomposition and calibration approach. Integrating these designs into a U-shaped convolutional backbone results in our efficient and effective network, termed UIRNet, for image restoration. Extensive experiments across single-degradation, all-in-one, and composite degradation restoration tasks validate the effectiveness of our method. Notably, UIRNet achieves state-of-the-art performance on 11 benchmark datasets covering six single-degradation scenarios and performs favorably against leading all-in-one approaches under a five-degradation setting. Importantly, our method also demonstrates robust performance on two challenging composite degradation image restoration benchmarks, as well as in ultra-high-definition (UHD) applications.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15492-15506"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097185","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Certified Robustness for Hashing via Randomized Smoothing 基于随机平滑的哈希证明鲁棒性

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-09 DOI: 10.1109/TCSVT.2026.3701674

Haomin Ni;Junyang Chen;Beibei Guan;Yan Pan;Jian Yin;Hanjiang Lai

{"title":"Certified Robustness for Hashing via Randomized Smoothing","authors":"Haomin Ni;Junyang Chen;Beibei Guan;Yan Pan;Jian Yin;Hanjiang Lai","doi":"10.1109/TCSVT.2026.3701674","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3701674","url":null,"abstract":"Deep hashing has garnered significant attention in recent years and has been successfully applied to a wide range of computer vision tasks, such as image retrieval. Despite its advantages, deep hashing methods are inherently vulnerable to adversarial attacks, which can easily manipulate the learned hash codes with imperceptible perturbations. To defend against the attacks, a key challenge lies in ensuring the stability of all hash bits to avoid global distortions in retrieval ranking. Although some defense strategies have been proposed to mitigate such vulnerabilities, they rely heavily on empirical robustness and lack provable or certifiable guarantees. In this paper, we propose a novel Certified Hashing based on randomized smoothing, which is a provably robust hashing model for adversarial perturbations. First, for each local bit, Certified Hashing applies Gaussian noise to the image and performs majority voting to obtain the most probable hash value. Within an <inline-formula> <tex-math>$ell _{2} $ </tex-math></inline-formula> ball of computable radius, each bit of the smoothed binary code can be theoretically proven to be consistent with the original one. Second, for the global hash code, we find that a few unstable bits can easily lead to the failure of the entire certification process, which limits the performance of Certified Hashing. Consequently, a bit selection and enhancement scheme is proposed to select stable bits based on the probability distribution estimated from Monte Carlo sampling. Comprehensive experiments on ImageNet, MS-COCO and NUS-WIDE datasets show the effectiveness of Certified Hashing in image retrieval application.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14929-14941"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097309","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

From Mapping to Composing: A Two-Stage Framework for Zero-Shot Composed Image Retrieval 从映射到合成：零镜头合成图像检索的两阶段框架

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-07-03 DOI: 10.1109/TCSVT.2026.3709874

Yabing Wang;Zhuotao Tian;Qingpei Guo;Zheng Qin;Sanping Zhou;Ming Yang;Nicu Sebe;Le Wang

{"title":"From Mapping to Composing: A Two-Stage Framework for Zero-Shot Composed Image Retrieval","authors":"Yabing Wang;Zhuotao Tian;Qingpei Guo;Zheng Qin;Sanping Zhou;Ming Yang;Nicu Sebe;Le Wang","doi":"10.1109/TCSVT.2026.3709874","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3709874","url":null,"abstract":"Composed Image Retrieval (CIR) is a challenging multimodal task that retrieves a target image based on a reference image and accompanying modification text. Due to the high cost of annotating CIR triplet datasets, zero-shot (ZS) CIR has gained traction as a promising alternative. Existing studies mainly focus on projection-based methods, which map an image to a single pseudo-word token. However, these methods face three critical challenges: 1) insufficient pseudo-word token representation capacity; 2) discrepancies between training and inference phases; and 3) reliance on large-scale synthetic data. To address these issues, we propose a two-stage framework where the training is accomplished from mapping to composing. In the first stage, we enhance image-to-pseudo-word token learning by introducing a visual semantic injection module and a soft text alignment objective, enabling the token to capture richer and fine-grained image information. In the second stage, we optimize the text encoder using a small amount of synthetic triplet data, enabling it to effectively extract compositional semantics by combining pseudo-word tokens with modification text for accurate target image retrieval. The strong visual-to-pseudo mapping established in the first stage provides a solid foundation for the second stage, making our approach compatible with both high- and low-quality synthetic data, and capable of achieving significant performance gains with only a small amount of synthetic data. Extensive experiments were conducted on three public datasets, achieving superior performance compared to existing approaches. Code is available at <uri>https://github.com/LiJiaBei-7/TSCIR</uri>","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15598-15609"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097423","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhancing Hyperbolic Vision--Language Representations via Adaptive Apertures and Contrastive Entailment 通过自适应孔径和对比蕴涵增强双曲视觉语言表征

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-08 DOI: 10.1109/TCSVT.2026.3701206

Changli Wang;Fang Yin;Suo Gao;Rui Wu

{"title":"Enhancing Hyperbolic Vision--Language Representations via Adaptive Apertures and Contrastive Entailment","authors":"Changli Wang;Fang Yin;Suo Gao;Rui Wu","doi":"10.1109/TCSVT.2026.3701206","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3701206","url":null,"abstract":"Hyperbolic representations have shown great promise for modeling visual semantic hierarchies, owing to their exponential representational capacity and inherent ability to encode hierarchical geometric structures. However, the existing hyperbolic vision–language model, MERU, suffers from two fundamental limitations, which we identify and substantiate through both theoretical analysis and empirical evaluation. First, the half-aperture angles frequently exceed the valid domain of the <inline-formula> <tex-math>$arcsin $ </tex-math></inline-formula> function, causing the entailment cones to degenerate into hemispheres and constraining the model’s representational expressiveness. Second, the entailment loss considers only positive samples, lacking explicit constraints to exclude negative examples; empirically, we observe that approximately 50% of negative samples fall within each entailment cone. To overcome these issues, we propose HyperConeNet, a novel framework that integrates a contrastive learning module, an entailment module, and an aperture diversity module. HyperConeNet decouples half-aperture prediction from curvature, reformulating it as a learnable parameterization, and incorporates both positive and negative samples into the entailment loss to strengthen the model’s discriminative power. Extensive experiments on 14 image classification and retrieval benchmarks demonstrate that HyperConeNet consistently outperforms baseline models while maintaining geometrically valid cone structures. Moreover, the predicted half-aperture angles remain strictly within the valid range <inline-formula> <tex-math>$left ({ 0, frac {pi }{2}}right)$ </tex-math></inline-formula>, ensuring stable and interpretable geometry. Further empirical analyses confirm that no negative samples fall within any entailment cone, indicating that HyperConeNet effectively addresses the two principal limitations of MERU. The code will be made publicly available soon.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15523-15538"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

S2-Net: Synergizing Spectral Compensation and Spatial Structure for Light Field Reconstruction S2-Net：协同光谱补偿和空间结构的光场重建

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-07-02 DOI: 10.1109/TCSVT.2026.3709274

Baoshuai Wang;Ping An;Yifan Mao;Yilei Chen;Xinpeng Huang;Chao Yang;Ce Zhu

{"title":"S2-Net: Synergizing Spectral Compensation and Spatial Structure for Light Field Reconstruction","authors":"Baoshuai Wang;Ping An;Yifan Mao;Yilei Chen;Xinpeng Huang;Chao Yang;Ce Zhu","doi":"10.1109/TCSVT.2026.3709274","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3709274","url":null,"abstract":"Due to limitations in optical design and acquisition cost, densely sampled light fields (LFs) cannot be directly captured in practice, motivating LF angular reconstruction from sparsely sampled views. Reconstructing densely-sampled LFs from sparse views necessitates simultaneously modeling long-range epipolar geometry and preserving fine-grained textures. However, we identify a fundamental spectral bias in existing State-of-the-Art (SOTA) Transformer-based approaches: while global attention mechanisms excel at capturing large-disparity dependencies, they inherently act as low-pass filters, causing spectral collapse in high-frequency regions (e.g., textures and edges). To bridge this gap between global geometric consistency and local spectral fidelity, we propose a novel dual-domain calibration framework termed S2-Net, which synergizes Spectral compensation and Spatial structure. Specifically, we introduce a Spectral Residual Modulation (SRM) module that constructs an adaptive band-pass filter in the frequency domain and reverses the spectral attenuation caused by global attention via a learnable spectral mask. Furthermore, we introduce a Structural Prior Guidance (SPG) module in the spatial domain to inject an explicit structural prior, where a gating mechanism adaptively regulates feature propagation, suppressing texture over-smoothing while preserving geometric consistency in the feature space. Extensive experiments on synthetic and real-world LF datasets demonstrate that the proposed method consistently outperforms SOTA approaches in both reconstruction accuracy and visual quality.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15433-15447"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097542","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

DEADet: Detail Enhancement With Multi-Level Feature Joint Perception for Aircraft Fine-Grained Object Detection 基于多层次特征联合感知的飞机细粒度目标检测细节增强

IF 10.8 1区工程技术

IEEE Transactions on Circuits and Systems for Video Technology Pub Date : 2026-10-01 Epub Date: 2026-06-15 DOI: 10.1109/TCSVT.2026.3703851

Fulin Luo;Yan Guan;Chuan Fu;Tan Guo;Yong Li;Bo Du;Liangpei Zhang

{"title":"DEADet: Detail Enhancement With Multi-Level Feature Joint Perception for Aircraft Fine-Grained Object Detection","authors":"Fulin Luo;Yan Guan;Chuan Fu;Tan Guo;Yong Li;Bo Du;Liangpei Zhang","doi":"10.1109/TCSVT.2026.3703851","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3703851","url":null,"abstract":"Fine-grained aircraft detection in remote sensing images plays a crucial role in the field of military security, providing critical technical support for military reconnaissance, target tracking, and air defense early warning. However, the task suffers from both inadequate preservation of local features, which causes a loss of discriminative details, and a reliance on layer-specific feature interactions that fails to leverage cross-scale complementary information. To address these challenges, we propose a Detail Enhancement Aircraft Fine-grained Object Detection (DEADet), which is an efficient one-stage detector based on multi-level feature joint perception. First, to preserve fine-grained details, we design the Detail-Enhanced Feature Representation (DEFR) module based on a compress-enhance-reconstruct principle. Second, to achieve comprehensive cross-level feature perception and enhancement, we introduce the Multi-level Joint Feature Refiner (MJFR), which consists of a Multi-level Feature Fusion (MLFF) module and a Multi-level Feature Enhance (MLFE) module. Third, we propose the Normalized Angle and Center Loss (NAC Loss), incorporating both angle penalty and center penalty to improve orientation estimation and spatial localization, thereby guiding the model toward more precise feature learning. Extensive experiments on the FAIR1M-aircraft and MAR20 datasets demonstrate that our method achieves state-of-the-art performance with 88.87% and 85.51% mAP respectively, while maintaining high efficiency with only 23.07M parameters and 84.19G FLOPs, providing an effective solution for military remote sensing applications. The code is available at <uri>https://github.com/mysort/DEADet</uri>","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14958-14973"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097639","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0