IEEE Transactions on Circuits and Systems for Video Technology最新文献_第8页

HypTM: Efficient Vision Transformers With Hyperbolic Token Merging HypTM：具有双曲令牌合并的高效视觉变压器

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: 2026-07-01 DOI: 10.1109/TCSVT.2026.3709249

Xiaoyu Chen;Xin Zhang;Weiying Xie;Jitao Ma;Yunsong Li;Leyuan Fang

{"title":"HypTM: Efficient Vision Transformers With Hyperbolic Token Merging","authors":"Xiaoyu Chen;Xin Zhang;Weiying Xie;Jitao Ma;Yunsong Li;Leyuan Fang","doi":"10.1109/TCSVT.2026.3709249","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3709249","url":null,"abstract":"Vision Transformers (ViTs) have shown tremendous potential in various computer vision applications attribute to the attention mechanism, which comes at the expense of increased computational costs. Therefore, token compression methods such as pruning and merging are proposed to facilitate the deployment of ViTs on resource-constrained devices. However, existing approaches assess tokens in Euclidean space, neglecting their complex geometric and hierarchical relationships. In this paper, we propose a novel Hyperbolic Token Merging (HypTM) method by exploring the intrinsic properties of images in hyperbolic space. Our HypTM is inspired by the insight that the contribution of distinct tokens to the overall image varies across the hierarchy. To this end, we develop a hyperbolic-driven token assessment algorithm that dynamically adjusts token significance according to hierarchical structure. Additionally, we design a hierarchy enhancement merging strategy to advance correlations among tokens within the same hierarchy. Without any retraining, our HypTM achieves notable improvements over the state-of-the-arts (SOTAs), while maintaining comparable FLOPs (floating point operations). For the DeiT-S model on ImageNet dataset, we achieve a 41.3% reduction in FLOPs with only a 0.48% decrease in Top-1 accuracy, which is even 0.09% higher than that of the meticulously fine-tuned comparison approach.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14786-14799"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097122","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

Attention-Disentangled Uniform Orthogonal Feature Space Optimization for Few-Shot Object Detection 小镜头目标检测的注意力解纠缠均匀正交特征空间优化

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.2026.3710023

Taijin Zhao;Heqian Qiu;Yu Dai;Lanxiao Wang;Fanman Meng;Qingbo Wu;Hongliang Li

{"title":"Attention-Disentangled Uniform Orthogonal Feature Space Optimization for Few-Shot Object Detection","authors":"Taijin Zhao;Heqian Qiu;Yu Dai;Lanxiao Wang;Fanman Meng;Qingbo Wu;Hongliang Li","doi":"10.1109/TCSVT.2026.3710023","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3710023","url":null,"abstract":"Few-Shot Object Detection (FSOD) aims to detect objects with limited samples for novel classes, while relying on abundant data for base classes. Existing FSOD approaches, predominantly built on the Faster R-CNN detector, entangle objectness recognition and foreground classification within shared feature spaces. This paradigm inherently establishes class-specific objectness criteria and suffers from unrepresentative novel class samples. To resolve this limitation, we propose a Uniform Orthogonal Feature Space (UOFS) optimization framework. First, UOFS decouples the feature space into two orthogonal components, where magnitude encodes objectness and angle encodes classification. This decoupling enables transferring class-agnostic objectness knowledge from base classes to novel classes. Moreover, implementing the disentanglement requires careful attention to two challenges: (1) Base set images contain unlabeled foreground instances, causing confusion between potential novel class instances and backgrounds. (2) Angular optimization depends exclusively on base class foreground instances, inducing overfitting of angular distributions to base classes. To address these challenges, we propose a Hybrid Background Optimization (HBO) strategy: (1) Constructing a pure background base set by removing unlabeled instances in original images to provide unbiased magnitude-based objectness supervision. (2) Incorporating unlabeled foreground instances in the original base set into angular optimization to enhance distribution uniformity. Additionally, we propose a Spatial-wise Attention Disentanglement and Association (SADA) module to address task conflicts between class-agnostic and class-specific tasks. Experiments demonstrate that our method significantly outperforms existing approaches based on entangled feature spaces.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14565-14577"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097125","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

Exploring Cross-Domain Transferable Attacks via Decoupling Features 通过解耦特征探索跨域可转移攻击

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: 2026-06-18 DOI: 10.1109/TCSVT.2026.3705137

Lifeng Huang;Wenyong Feng;Han Wang;Haoxing Liu;Chen Wan;Qiong Huang

{"title":"Exploring Cross-Domain Transferable Attacks via Decoupling Features","authors":"Lifeng Huang;Wenyong Feng;Han Wang;Haoxing Liu;Chen Wan;Qiong Huang","doi":"10.1109/TCSVT.2026.3705137","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3705137","url":null,"abstract":"Most existing transferable adversarial attacks are fundamentally flawed because they assume attackers can access the training data of victim models, a premise that contradicts practical black-box scenarios. This issue is compounded in cross-domain attack setups, where inherent data distribution discrepancies hinder attack effectiveness by causing surrogate models to capture overly domain-specific features. To address these issues, we propose the Domain Decoupling Adversarial Attack (D2A), which enhances cross-domain transferability through two key perspectives: surrogate model disentanglement and adversarial perturbation optimization. Firstly, we decouple the pre-trained surrogate model into two models: coarse-grained and fine-grained domain-aware models. The former one is developed through model enhancement, specifically by filtering out domain-specific information with additional modules. Concurrently, the latter one is built via data augmentation, simulating diverse data distributions to fine-tune the original surrogate model. This enables each decoupled model to focus on capturing either coarse-grained or fine-grained features, respectively. Secondly, we introduce a dual-level perturbation optimization strategy. It jointly perturbs global and local features within both the coarse-grained and fine-grained models, thereby guiding the generator to produce adversarial examples with improved cross-domain transferability. Extensive experiments conducted on various datasets demonstrate that D2A achieves an average attack success rate improvement of 8.0% compared to SOTA transferable attacks, with particularly notable gains of 12.28% in fine-grained domains. Furthermore, D2A also exhibits substantial benefits in cross-architecture attack scenarios, further validating the effectiveness and robustness of the proposed method. The code is available at <uri>https://github.com/sdfdsdff/D2A</uri>.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15745-15760"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097399","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

LR-Depth: Enhancing Self-Supervised Monocular Depth Estimation in Low-Resolution Scenarios LR-Depth：在低分辨率场景下增强自监督单目深度估计

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.2026.3703327

Xiao Ke;Chaoying Chen;Jingyi Fang;Huanqi Wu;Yuzhen Niu

{"title":"LR-Depth: Enhancing Self-Supervised Monocular Depth Estimation in Low-Resolution Scenarios","authors":"Xiao Ke;Chaoying Chen;Jingyi Fang;Huanqi Wu;Yuzhen Niu","doi":"10.1109/TCSVT.2026.3703327","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3703327","url":null,"abstract":"Existing self-supervised monocular depth estimation methods have achieved impressive results by reducing reliance on expensive labeled data. However, their performance degrades significantly in low-resolution scenes due to challenges like missing edge details, difficulty in modeling blur regions, and inaccuracies in upsampling geometric structure reconstruction. To address this issue, we propose LR-Depth, an efficient self-supervised monocular depth estimation framework tailored for low-resolution scenes. Firstly, we propose a multi-scale wavelet edge enhancement that captures high-frequency edge information to improve depth estimation in edge regions. Meanwhile, to improve blurred regions, we leverage wavelet blur mapping to refine low-frequency component representation and enhance the modeling accuracy of depth distributions. Finally, we introduce geometric reconstruction upsampling that integrates position initialization with local offset constraints, enabling precise recovery of complex geometric structures while ensuring geometric consistency of the depth map. Extensive experiments on the KITTI dataset demonstrate that LR-Depth achieves state-of-the-art performance in low-resolution self-supervised monocular depth estimation, particularly excelling in edge clarity and depth prediction for blurred regions. Additionally, generalizability experiments on the Make3D dataset further confirm the framework’s robustness and broad applicability.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14726-14739"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097504","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

Toward Low-Latency Event Stream-Based Visual Object Tracking: A Slow-Fast Approach 基于低延迟事件流的视觉对象跟踪：一种慢-快方法

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: 2026-06-26 DOI: 10.1109/TCSVT.2026.3707744

Shiao Wang;Xiao Wang;Liye Jin;Bo Jiang;Lin Zhu;Lan Chen;Yonghong Tian;Bin Luo

{"title":"Toward Low-Latency Event Stream-Based Visual Object Tracking: A Slow-Fast Approach","authors":"Shiao Wang;Xiao Wang;Liye Jin;Bo Jiang;Lin Zhu;Lan Chen;Yonghong Tian;Bin Luo","doi":"10.1109/TCSVT.2026.3707744","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3707744","url":null,"abstract":"Existing tracking algorithms typically rely on low-frame-rate RGB cameras coupled with computationally intensive deep neural network architectures to achieve effective tracking. However, such frame-based methods inherently face challenges in achieving low-latency performance and often fail in resource-constrained environments. Visual object tracking using bio-inspired event cameras has emerged as a promising research direction in recent years, offering distinct advantages for low-latency applications. In this paper, we propose a novel Slow-Fast tracking paradigm that flexibly adapts to different operational requirements, termed SFTrack. The proposed framework supports two complementary modes, i.e., a high-precision slow tracker for scenarios with sufficient computational resources, and an efficient fast tracker tailored for latency-aware, resource-constrained environments. Specifically, our framework first performs graph-based representation learning from high-temporal-resolution event streams, and then integrates the learned graph-structured information into two FlashAttention-based vision backbones, yielding the slow and fast trackers, respectively. The fast tracker achieves low latency through a lightweight network design and by producing multiple bounding box outputs in a single forward pass. Finally, we seamlessly combine both trackers via supervised fine-tuning and further enhance the fast tracker’s performance through a knowledge distillation strategy. Extensive experiments on public benchmarks, including FE240, COESOT, and EventVOT, demonstrate the effectiveness and efficiency of our proposed method across different real-world scenarios. The source code has been released on <uri>https://github.com/Event-AHU/SlowFast_Event_Track</uri>","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"16051-16064"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097546","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

ReAlign: Re-Aligning Intra-Class Misalignment for Few-Shot Segmentation ReAlign：重新对准类内不对齐的镜头分割

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: 2026-06-18 DOI: 10.1109/TCSVT.2026.3705224

Jiaxiang Fang;Shiqiang Ma;Guihua Duan;Fei Guo;Shengfeng He

{"title":"ReAlign: Re-Aligning Intra-Class Misalignment for Few-Shot Segmentation","authors":"Jiaxiang Fang;Shiqiang Ma;Guihua Duan;Fei Guo;Shengfeng He","doi":"10.1109/TCSVT.2026.3705224","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3705224","url":null,"abstract":"Few-shot segmentation aims to segment unseen query objects using a few annotated support samples, where robustness against intra-class misalignment between support and query features is crucial. We introduce <italic>ReAlign</i>, a method that corrects these discrepancies by realigning intra-class misalignments. Our approach is underpinned by two key observations. First, intra-class consistent knowledge and spatial association knowledge effectively mitigate class inconsistencies. To leverage these insights, we developed the Harmonized Debiasing Module, which employs multimodal approaches to extract this essential knowledge. Addressing the potential biases from limited sample sizes, this module incorporates multi-teacher distillation and compression techniques to enhance representation robustness. Second, we observe that query features with strong semantic signals exhibit enhanced self-matching properties, which help reduce intra-class misalignment. Additionally, exploiting the rich textual knowledge from visual-language pre-trained models optimizes alignment. Building on these, we propose the High-CLIP method, which refines existing vision-language pre-training frameworks by incorporating text features to extract high-confidence query features for autonomous decision-making, further reducing misalignment issues. <italic>ReAlign</i> significantly outperforms existing methods, setting new benchmarks. It achieves mIoU scores of 71.6% and 76.4% for 1-shot and 5-shot segmentation, respectively, on PASCAL-<inline-formula> <tex-math>$5^{i}$ </tex-math></inline-formula>.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"14991-15006"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097644","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

FedHarmony: A Unified Framework for Addressing Clinical Heterogeneity in Federated Medical Image Analysis FedHarmony：解决联邦医学图像分析中临床异质性的统一框架

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: 2026-07-02 DOI: 10.1109/TCSVT.2026.3709668

Haoyu Mao;Tengfei Cao;Yuli Liu;Han Xiao;Changqiao Xu

{"title":"FedHarmony: A Unified Framework for Addressing Clinical Heterogeneity in Federated Medical Image Analysis","authors":"Haoyu Mao;Tengfei Cao;Yuli Liu;Han Xiao;Changqiao Xu","doi":"10.1109/TCSVT.2026.3709668","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3709668","url":null,"abstract":"Federated Learning (FL) offers a privacy-preserving approach for collaborative modeling of federated medical imaging data across multiple centers. However, clinical heterogeneity, which arises from differences in imaging equipment, scanning protocols, and patient populations across centers, leads to performance drift and limits the model’s global generalization. Existing methods typically optimize client selection and model aggregation separately, which makes it difficult to mitigate the combined negative effects of unrepresentative training queues and unstable model fusion. In this study, we propose FedHarmony, a unified framework designed to address clinical heterogeneity in federated medical image analysis. The focus of this work is heterogeneity-aware federated learning for multi-center medical imaging, where variations in imaging quality, acquisition conditions, and patient distributions may jointly impair robustness, fairness, and generalization. Specifically, we introduce two novel mechanisms: Proactive Client Orchestration (PCO) and Stability-aware Model Fusion (SMF). PCO strategically guides training by evaluating each client’s potential performance gains and gradient complementarity, thereby improving overall effectiveness. SMF enhances robustness by introducing noise perturbations during local training to promote flatter minima, and it assigns aggregation weights according to model stability, thus improving both generalization and training stability. Experimental results show that FedHarmony outperforms existing state-of-the-art methods in accuracy and stability across diverse medical imaging tasks, while effectively balancing communication costs and model performance. Ablation studies further confirm the complementary benefits of PCO and SMF.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15962-15976"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097567","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

Robust Coverless Information Hiding With Object-Presence Hashing and Vision Transformer-U-Net-Based Reconstruction 基于对象存在哈希的鲁棒无覆盖信息隐藏和基于视觉变换- u - net的重构

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: 2026-06-04 DOI: 10.1109/TCSVT.2026.3700219

Sangeeta Gautam;Manoj Kumar

{"title":"Robust Coverless Information Hiding With Object-Presence Hashing and Vision Transformer-U-Net-Based Reconstruction","authors":"Sangeeta Gautam;Manoj Kumar","doi":"10.1109/TCSVT.2026.3700219","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3700219","url":null,"abstract":"Covert communication in visual media, which is essential for advanced intelligent systems, faces persistent challenges due to the susceptibility of conventional steganography to steganalysis and the limited robustness and payload capacity of existing coverless information hiding methods, particularly under geometric distortions. To address these challenges, a coverless information hiding framework is proposed that combines robust block-based object-presence hash sequence generation with a receiver-side preprocessing pipeline for geometric distortion detection, inverse geometric correction (applied only when the corresponding transformation parameters can be reliably estimated), and image restoration using a Vision Transformer-U-Net–based reconstruction network with selective masking. This design enables accurate spatial realignment and effective reconstruction of degraded regions, which is critical for reliable hash regeneration after transmission. Unlike existing coverless methods that operate with fixed short hash lengths (e.g., 15 bits, typically requiring 548-8193 images to transmit 1 KB of data), the proposed framework supports scalable hash lengths from 8 to 64 bits per image. With a 36-bit hash length, the number of images required to transmit 1 KB of data is reduced to approximately 228, corresponding to an efficiency improvement of about 50-60% under standard 15-bit hashing settings. Comprehensive experiments on the INRIA Holidays, Caltech-256, PASCAL VOC2012, MS COCO2014 and ImageNet datasets demonstrate near-perfect retrieval accuracy (up to 100%) under a wide range of geometric and statistical attacks, confirming the reliability and practicality of the proposed framework for covert communication in real-world visual environments.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15652-15668"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097695","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

Memristive Coupled Dual Rulkov Neuron Model With Multi-Type Hyperchaotic Firings and Application to Image Encryption for IoMT 具有多类型超混沌激发的记忆耦合双Rulkov神经元模型及其在IoMT图像加密中的应用

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: 2026-06-26 DOI: 10.1109/TCSVT.2026.3707598

Xiaohan Zhang;Sen Zhang;Siyao Lu;Xiaoping Wang;Zhigang Zeng

{"title":"Memristive Coupled Dual Rulkov Neuron Model With Multi-Type Hyperchaotic Firings and Application to Image Encryption for IoMT","authors":"Xiaohan Zhang;Sen Zhang;Siyao Lu;Xiaoping Wang;Zhigang Zeng","doi":"10.1109/TCSVT.2026.3707598","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3707598","url":null,"abstract":"Memristors are regarded as excellent carriers for simulating biological synapses due to their unique memory properties. When employing memristive neuron models to simulate biological neuronal firing behaviors, the complexity of firing activities is a critical measure in assessing model performance. However, the diversity and complexity of firing patterns generated by individual neurons or memristor-coupled single neurons typically suffer from certain limitations. To solve this problem, this paper employs a universal discrete memristor for synaptic coupling between two Rulkov neurons to build a memristive coupled dual Rulkov neuron model (MCDRNM). By switching the memristor’s internal parameters, the MCDRNM can generate a controllable number of multi-type hyperchaotic firing patterns, including multi-foot-shaped and multi-pigeon-shaped patterns. Through adjustments of the initial conditions of the memristor, the MCDRNM not only exhibits heterogeneous firing multistability but also induces homogeneous firing multistability with infinitely coexisting single-foot-shaped/pigeon-shaped hyperchaotic firing patterns. Furthermore, firing pattern transitions are also revealed in this MCDRNM. Meanwhile, STM32 hardware experiments are conducted, demonstrating the physical realizability of the model. Finally, based on the hyperchaotic firing sequences of the MCDRNM, an efficient image encryption method suitable for the Internet of Medical Things (IoMT) is designed. The evaluation results demonstrate that this encryption scheme is as efficient as it is secure. It achieves high encryption efficiency while maintaining strong resistance to attacks.","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15159-15171"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097836","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

DifferAD-R1: A Difference-Guided Industrial Anomaly Localization With Multimodal Large Language Models 差分- r1：基于多模态大语言模型的差分导向工业异常定位

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: 2026-06-18 DOI: 10.1109/TCSVT.2026.3704942

Dingrong Wang;Xian Tao;Zhen Qu;Hengliang Luo;Xinyi Gong;Fei Shen;Zhengtao Zhang;Guiguang Ding

{"title":"DifferAD-R1: A Difference-Guided Industrial Anomaly Localization With Multimodal Large Language Models","authors":"Dingrong Wang;Xian Tao;Zhen Qu;Hengliang Luo;Xinyi Gong;Fei Shen;Zhengtao Zhang;Guiguang Ding","doi":"10.1109/TCSVT.2026.3704942","DOIUrl":"https://doi.org/10.1109/TCSVT.2026.3704942","url":null,"abstract":"Industrial anomaly localization aims to accurately identify and localize abnormal regions in industrial products, addressing the critical challenge of detecting unseen defect categories in real-world scenarios. Traditional closed-set methods often suffer from poor cross-scenario generalization, while existing Multimodal Large Language Model (MLLM)-based approaches face two core limitations: they either adopt QA-style paradigms misaligned with the practical demands of localization, or rely on standard optimization techniques such as Group Relative Policy Optimization (GRPO), which fails to deliver effective learning signals for subtle defects. To tackle these issues, this paper proposes DifferAD-R1, an MLLM-augmented reinforcement learning framework tailored for industrial anomaly localization. We design a Difference-Guided dual-image paradigm, which reformulates the localization task as a one-shot difference grounding problem to effectively explore cross-scenario anomalies. A Dual-Consistency Localization Reward (DCLR) is developed for hard-to-detect anomalies, enhancing optimization stability and robustness. Additionally, we integrate a difficulty-aware strategy with adaptive reweighting and group-wise resampling to prioritize learning on challenging instances. To facilitate evaluations in real-world industrial settings, we construct the AD-DualDiff dataset, comprising 13K paired images across 20 categories. Experimental results demonstrate that DifferAD-R1 significantly outperforms existing baselines and achieves competitive performance compared to large-scale models like Qwen3-VL (235B parameters). Our code is publicly available at: <uri>https://github.com/Rong2026/work-1</uri>","PeriodicalId":13082,"journal":{"name":"IEEE Transactions on Circuits and Systems for Video Technology","volume":"36 10","pages":"15007-15022"},"PeriodicalIF":10.8,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097897","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