Unsupervised Domain Adaptation (UDA) has shown remarkable success in medical image segmentation but its practical deployment is often constrained by strict privacy regulations that prohibit access to source-domain data. Source-Free Domain Adaptation (SFDA) addresses this limitation by enabling knowledge transfer from pre-trained source models without requiring private source data. Existing SFDA methods for medical image segmentation mainly rely on self-training with pseudo-labels generated by the source model. However, severe domain discrepancies introduce substantial pseudo-label noise, while non-uniform domain shifts across target samples lead to heterogeneous data distributions, rendering uniform adaptation strategies suboptimal. To address these challenges, we propose Entropy-aware Mutual Student Co-training (EMSC), a novel SFDA framework comprising one teacher model and two student models. An Entropy-Guided Difficulty Assessment (EGDA) module partitions target samples into source-similar and source-dissimilar subsets based on predictive uncertainty. Each subset is handled by an independent student branch equipped with a Subset-specific Adaptive Injection (SAI) module, which injects semantic anchors for source-similar samples and structural noise for source-dissimilar samples to enable differentiated adaptation and suppress pseudo-label noise. Extensive experiments on two widely used medical image segmentation benchmarks demonstrate that EMSC consistently outperforms state-of-the-art SFDA methods across multiple evaluation metrics, validating its robustness under heterogeneous domain shifts.

Depression has become a critical global public health challenge, creating an urgent need for automated and scalable screening solutions. Social media platforms, which capture rich and spontaneous multi-modal behavioral data, offer a promising avenue for detecting early signs of mental distress. However, existing depression detection methods predominantly rely on static multi-modal fusion strategies and frequently fail to effectively tackle cross-modal semantic gaps. To address these limitations, we propose a Mutual Guidance Multi-modal Network with Complementary Graph Learning (MGMN) for depression detection by observing individuals' behavioral performance on social media. Specifically, a cross-modal mutual guidance mechanism is designed to dynamically construct a complementary graph by using mutual similarities within and across visual and acoustic modalities common in social media. More specifically, based on this complementary graph, a modality-specific adaptive residual learning module is applied to each modality to stabilize deep feature learning and preserve modality-specific and complementary information via graph-conditioned adaptive residual fusion. Furthermore, the refined uni-modal features are subsequently fed into a joint-modal fusion and prediction module to output the final disease prediction probability. Extensive experiments on the MUD3, LMVD, and D-vlog datasets demonstrate our proposed method's superiority over state-of-the-art methods, confirming that the proposed framework provides a robust and effective solution for mental health monitoring. Codes are available at https://github.com/Petofi-romance/MGMN

ABSTRACT Diffusion‐based unsupervised anomaly detection in medical images has emerged as an effective paradigm, leveraging unlabelled healthy data to precisely characterize the distribution of normal anatomy and identify a wide range of pathological abnormalities. The method reconstructs a pseudo‐healthy image from a potentially anomalous input and identifies anomalies by measuring pixel‐wise reconstruction errors. However, existing approaches often preserve anomalous regions in the reconstruction, resulting in less prominent anomaly segmentation. Additionally, their inability to accurately restore normal areas can lead to increased false positives. In this work, we propose CS‐Unet to advance this paradigm by realizing the concept of anomaly‐aware reconstruction, defined as reconstructions that are consciously devoid of anomalies while faithfully restoring normal regions. Firstly, we propose a compression‐expansion DenseNet (CompExDenseNet), which performs a dense cascade of nonlinear dimension transformations to extract compact feature representations, suppressing the reconstruction of anomalous patterns. Secondly, we design an attention gate (AG) unit to control the flow of low‐frequency information, mitigating the leakage of anomalous information. Finally, we propose a frequency‐domain adaptive residual convolution (FreAR) module that selectively enhances the most relevant frequency components to facilitate high‐fidelity restoration of normal regions. Experimental results demonstrate that CS‐Unet achieves outstanding performance in unsupervised anomaly detection, confirming its effectiveness.

Fine-grained image clustering (FGIC) is a highly challenging task due to the large intra-class variance, small inter-class variance, and lack of annotation, aiming at grouping images into fine-grained subcategories. Existing FGIC methods generally learn parameterized localization networks to capture key objects for better clustering performance. Despite yielding promising improvements, these methods still have limitations. First, localization networks introduce additional parameters, and not all localized regions are beneficial for clustering. Second, FGIC requires more detailed semantic descriptions, however, current methods only mine supervisory signals from images, making it difficult to meet practical demands. For addressing these limitations, this paper proposes the FGCLIP-based augmented language-driven contrastive clustering (FGCLIP-ALCC) network, which uses frozen FG-CLIP to introduce external knowledge and designs parameter-efficient text branch to accurately locate key objects and learn fine-grained text semantics. More specifically, FGCLIP-ALCC contains two components: the augmented language-driven fine-grained semantic learner (ALFSL) and the multi-modal contrastive clustering heads (MmCCH). First, the ALFSL is designed with a three-stream architecture to enhance robustness, ensure the diversity and accuracy of text descriptions generated subsequently, and utilize image branches to extract visual features. Next, the text branch in ALFSL uses an augmentation-driven diverse text generation module to generate coarse-grained text descriptions, a text fine-graining module to capture key object semantics and refine the text descriptions, and a text filtering module along with a text fusion operation to enhance the intra-class cohesion of text semantics and (to) obtain unique fine-grained text embeddings for each image. Finally, the MmCCH is used to inject text semantics into visual features and obtain clustering results. Experimental results on five fine-grained datasets and four coarse-grained datasets show that FGCLIP-ALCC outperforms state-of-the-art clustering methods on all datasets and metrics, while requiring only 1.2M additional learnable parameters. The code will be released at https://github.com/xjq425/FGCLIP-ALCC.

With the proliferation of multi-modal data, safe and efficient multi-modal hashing retrieval has become a pressing research challenge, particularly due to concerns over data privacy during centralized processing. To address this, we propose Prototype-based Federated Multi-modal Hashing (PFMH), an innovative framework that seamlessly integrates federated learning with multi-modal hashing techniques. PFMH achieves fine-grained fusion of heterogeneous multi-modal data, enhancing retrieval accuracy while ensuring data privacy through prototype-based communication, thereby reducing communication costs and mitigating risks of data leakage. Furthermore, using a prototype completion strategy, PFMH tackles class imbalance and statistical heterogeneity in multi-modal data, improving model generalization and performance across diverse data distributions. Extensive experiments demonstrate the efficiency and effectiveness of PFMH within the federated learning framework, enabling distributed training for secure and precise multi-modal retrieval in real-world scenarios.

Zero-shot anomaly detection aims to identify anomalies in unseen classes using knowledge learned from seen classes. Advances in vision-language models like CLIP have demonstrated strong potential for zero-shot anomaly detection tasks. However, CLIP was originally designed to align text with global visual features, focusing on image-level semantics information, making it difficult to capture local anomaly features in pixel-level localization tasks accurately. To address this issue, we propose a CLIP-based Adaptive and Static Prompts learning method for zero-shot anomaly detection (ASP-CLIP). It combines static text prompts and adaptive category prompts to form hybrid text prompts, optimizing text representations. Static text prompts capture general information across categories, while adaptive category prompts are generated by an image adapter module that introduces category-specific information by mapping visual features to the text embedding space. Specifically, we design the Stepwise Adjustment module that improves the alignment between local visual features and text embeddings by utilizing similarity maps identified from shallow local visual features to refine deep local visual features. Experimental results validate the effectiveness of our approach, demonstrating highly competitive performance on the MVTec and VisA datasets.

Subspace clustering contributes a lot to the development of unsupervised hyperspectral image (HSI) classification task due to its ability of processing high-dimensional data. Generally, subspace clustering suffers from the bottlenecks of high computational cost when dealing with large-scale HSI. Some researches address this problem by using superpixel segmentation. However, the introduction of superpixel segmentation may cause some adverse effects, including interference from non-target region samples and the oversmoothing of superpixel-level samples. To overcome these issues, we propose a masked superpixel contrastive subspace clustering (MSCSC) for large-scale HSI classification. Specifically, we leverage hyperspectral masked autoencoder rather than conventional autoencoder as the backbone to mitigate the interference of non-target region samples during feature extraction. Then, based on this backbone, we integrate contrastive learning with self-expressiveness based subspace clustering to form the superpixel-level contrastive subspace clustering network, which aims to learn the discriminative superpixel-level self-representation. Moreover, we design a novel data augmentation strategy to ensure learn global and local information, improving the distinguishability of superpixel-level features. Experiments on four popular HSI datasets validate the superiority of our proposed method, with a large accuracy improvement compared to state-of-the-art methods.

Multi-view Subspace Clustering (MVSC) effectively aggregating multiple data sources to promise clustering performance. Recently, various anchor-based variants have been introduced to effectively alleviate the computation complexity of MVSC. Although satisfactory advancement has been achieved, existing methods either independently learn anchor matrices and their anchor representations or learn a consensus anchor matrix and unified anchor representation, failing to capture both consistency and complementary information simultaneously. In addition, the time complexity of obtaining clustering results by applying Singular Value Decomposition (SVD) on the anchor representation matrix remains high. To tackle the above problems, we propose an Adaptive Anchor-guided Representation Learning for Efficient Multi-view Subspace Clustering (A2RL-EMVSC) framework, which integrates consensus anchors learning, anchor-guided representation learning and matrix factorization to enhance clustering performance and scalability. Technically, the proposed method learns view-specific anchor representation matrices by consensus anchors guidance, which simultaneously exploit consistency and complementary information. Moreover, by applying matrix decomposition to the view-specific anchor representation matrices, clustering results can be achieved with linear time complexity. Extensive experiments on ten challenging multi-view datasets show that the proposed method can improve the effectiveness and superiority of clustering compared with state-of-the-art methods.

Conventional domain adaptation (DA) for person re-identification (ReID) aims to bridge the domain gap but often requires direct use of fully labeled source and target domains, raising significant data privacy concerns due to the inclusion of personal identity information (PII) in raw data. Source-free domain adaptation (SFDA) for person ReID effectively preserves PII within the authorized source model. Nevertheless, these methods are vulnerable to data privacy (e.g., portrait rights) of the target domain during retrieval, where attackers can exploit pedestrian images for malicious generation, leading to damage to an individual’s reputation. Beyond these limitations, we propose a novel framework called SecureDA to address privacy-preserving SFDA for person ReID, which can generate a privacy key to defend against potential attacks on PII. Technically, we introduce domain-specific adversarial attacks into DA, where the protected query and gallery images are encrypted to ensure secure image retrieval. Furthermore, we employ two simultaneous processes: 1) The global–local adversarial pathway (GLAP) leverages encrypted and original images as adversarial pairs, thereby fostering the development of robust ReID models; 2) The global–local collaborative pathway (GLCP) is mastered through positive pairs collected from the same domain, effectively mitigating the pernicious catastrophic forgetting phenomenon. Extensive experiments show that SecureDA achieves state-of-the-art performance on multiple DA benchmarks and even outperforms the conventional DA and SFDA methods, which inherently compromise data privacy.

Conventional domain adaptive (DA) methods for person re-identification (ReID) face knowledge transfer challenges when labeled data from the source domain cannot be accessed due to privacy constraints. Although the methods operating under source-absent DA settings attempt to address this challenge by using models pretrained on the source domain in their mutual teaching frameworks, failing to capture domain divergence in scenarios in which the source data are completely inaccessible can simultaneously introduce issues related to mutual convergence. In response, we introduce an adversarial attacks-driven mutual teaching (AAMT) framework as an innovative and applicable source-free DA person ReID scheme. Specifically, we first carefully develop a perturbation generator to generate source-style adversarial examples by leveraging a pretrained source model. Then, these diverse adversarial examples are employed to attack the mutual teaching model, implicitly measuring the domain divergence. Accordingly, we design a contrastive learning loss to enlarge the differences between the training pairs and further mitigate the mutual convergence issue. Extensive experiments demonstrate that AAMT outperforms the existing methods under both conventional and source-absent DA settings, achieving state-of-the-art performance.

Domain adaption (DA) for person re-identification (ReID) has attained considerable progress by transferring knowledge from a source domain with labels to a target domain without labels. Nonetheless, most of the existing methods require access to source data, which raises privacy concerns. Source-free DA has recently emerged as a response to these privacy challenges, yet its direct application to open-set pedestrian re-identification tasks is hindered by the reliance on a shared category space in existing methods. Current source-free DA approaches for person ReID still encounter several obstacles, particularly the divergence-agnostic problem and the notable domain divergence due to the absent source data. In this article, we introduce an Instance-level Adversarial Mutual Teaching (IAMT) framework, which utilizes adversarial views to tackle the challenges mentioned above. Technically, we first elaborately develop a variance-based division (VBD) module to segregate the target data into instance-level subsets based on their similarity and dissimilarity to the source using the source-trained model, implicitly tackling the divergence-agnostic problem. To mitigate domain divergence, we additionally introduce a dynamic adversarial alignment (DAA) strategy, aiming to enhance the consistence of feature distribution across domains by employing adversarial instances from the target data to confuse the discriminators. Experiments reveal the superiority of the IAMT over state-of-the-art methods for DA person ReID tasks, while preserving the privacy of the source data.

Automatic segmentation of COVID-19 lesions is essential for computer-aided diagnosis. However, this task remains challenging because widely-used supervised based methods require large-scale annotated data that is difficult to obtain. Although an unsupervised method based on anomaly detection has shown promising results in [1], its performance is relatively poor. We address this problem by proposing a pixel-level and affinity-level knowledge distillation method. It obtains a pre-trained teacher network with rich semantic knowledge of CT images by constructing and training an auto-encoder at first, and then trains a student network with the same architecture as the teacher by distilling the teacher’s knowledge only from normal CT images, and finally localizes COVID-19 lesions using the feature discrepancy between the teacher and the student networks. Besides, except for the traditional pixel-level distillation, we design the affinity-level distillation that takes into account the pairwise relationship of features to fully distill effective knowledge. We evaluate this method by using three different COVID-19 datasets and the experimental results show that the segmentation performance is largely improved when it is compared with the other existing unsupervised anomaly detection methods.

Abstract Face detection is playing a pivotal role for crowd counting and abnormal events detection. However, it is vulnerable to face presentation attacks by printed photos, videos, and 3D masks of real human faces. Although numerous detection techniques based on deep learning have been employed to address the problem of face presentation attacks, there are still several weaknesses in these approaches, such as high algorithm complexity and a lack of detection ability. To overcome these weaknesses, a method based on a multilevel fusion network with Laplacian embedding (MFNet‐LE) for the detection of face presentation attacks is proposed. First, a shallow network that contains just three layers was developed, which makes the model faster. Then, an optimised multilevel fusion strategy was developed to combine the input with the output of all previous layers to improve the detection ability of the method. Finally, a Laplacian embedding algorithm is introduced to maintain the inter‐class discrimination and penalise the intra‐class distance. Under the joint supervision of Laplacian loss and softmax loss, the proposed approach can obtain more discriminative features, which enhance the accuracy of attack detection. Experiments were conducted with three public databases for face presentation attacks: CASIA FASD, Idiap Replay Attack database and MSU USSA. The results demonstrate that the MFNet‐LE model can outperform the state‐of‐the‐art methods.

Identity authentication based on face recognition has been significantly improved due to the outstanding ability of face detection, thus it plays an important role in society. However, face recognition system might be deceived by malicious face spoof attacks raising risk from both safety and property. The algorithm to accurately detect face anti-spoofing in identity authentication system is becoming crucial. In this paper, a shallow convolutional neural network with laplacian embedding (shallowCNN-LE) is proposed for face anti-spoofing. Two different types of features are concatenated to accurately detect the face liveness, including depth features and dynamic texture features. First, the developed shallow CNN model contains four layers which make the model faster. Second, we integrate dynamic texture features extracted by using the dual tree complex wavelet transform (DT-CWT) with the depth features as input features to feed into the proposed model. Finally, we propose a laplacian embedding algorithm, which can maintain the inter-class discrimination and penalize the distance of intra-class. When embedding the laplacian loss with the softmax loss, the proposed method can obtain much more discriminative features, which is helpful to detect face anti-spoofing. Experimental results on public databases of CASIA FASD, Replay attack and MSU USSA database demonstrate that our proposed method outperforms the state-of-the-art methods for face anti-spoofing detection.

Biometric applications have been used globally in everyday life. However, conventional biometrics is created and optimized for high-security scenarios. Being used in daily life by ordinary untrained people is a new challenge. Facing this challenge, designing a biometric system with prior constraints of ergonomics, we propose ergonomic biometrics design model, which attains the physiological factors, the psychological factors, and the conventional security characteristics. With this model, a novel hand-based biometric system, door knob hand recognition system (DKHRS), is proposed. DKHRS has the identical appearance of a conventional door knob, which is an optimum solution in both physiological factors and psychological factors. In this system, a hand image is captured by door knob imaging scheme, which is a tailored omnivision imaging structure and is optimized for this predetermined door knob appearance. Then features are extracted by local Gabor binary pattern histogram sequence method and classified by projective dictionary pair learning. In the experiment on a large data set including 12 000 images from 200 people, the proposed system achieves competitive recognition performance comparing with conventional biometrics like face and fingerprint recognition systems, with an equal error rate of 0.091%. This paper shows that a biometric system could be built with a reliable recognition performance under the ergonomic constraints.

Biometric recognition systems have been widely used globally. However, one effective and highly accurate biometric authentication method, palmprint recognition, has not been popularly applied as it should have been, which could be due to the lack of small, flexible and user-friendly acquisition systems. To expand the use of palmprint biometrics, we propose a novel palmprint acquisition system based on the line-scan image sensor. The proposed system consists of a customized and highly integrated line-scan sensor, a self-adaptive synchronizing unit, and a field-programmable gate array controller with a cross-platform interface. The volume of the proposed system is over 94% smaller than the volume of existing palmprint systems, without compromising its verification performance. The verification performance of the proposed system was tested on a database of 8000 samples collected from 250 people, and the equal error rate is 0.048%, which is comparable to the best area camera-based systems.