Monocular 3D Visual Grounding (Mono3DVG) is an emerging task that locates 3D objects in RGB images using text descriptions with geometric cues. However, existing methods face two key limitations. Firstly, they often over-rely on high-certainty keywords that explicitly identify the target object while neglecting critical spatial descriptions. Secondly, generalized textual features contain both 2D and 3D descriptive information, thereby capturing an additional dimension of details compared to singular 2D or 3D visual features. This characteristic leads to cross-dimensional interference when refining visual features under text guidance. To overcome these challenges, we propose Mono3DVG-EnSD, a novel framework that integrates two key components: the CLIP-Guided Lexical Certainty Adapter (CLIP-LCA) and the Dimension-Decoupled Module (D2M). The CLIP-LCA dynamically masks high-certainty keywords while retaining low-certainty implicit spatial descriptions, thereby forcing the model to develop a deeper understanding of spatial relationships in captions for object localization. Meanwhile, the D2M decouples dimension-specific (2D/3D) textual features from generalized textual features to guide corresponding visual features at same dimension, which mitigates cross-dimensional interference by ensuring dimensionally-consistent cross-modal interactions. Through comprehensive comparisons and ablation studies on the Mono3DRefer dataset, our method achieves state-of-the-art (SOTA) performance across all metrics. Notably, it improves the challenging Far([email protected]) scenario by a significant +13.54%.

Multimodal anomaly detection (MAD) aims to exploit both texture and spatial attributes to identify deviations from normal patterns in complex scenarios. However, zero-shot (ZS) settings arising from privacy concerns or confidentiality constraints present significant challenges to existing MAD methods. To address this issue, we introduce ZUMA, a training-free, Zero-shot Unified Multimodal Anomaly detection framework that unleashes CLIP's cross-modal potential to perform ZS MAD. To mitigate the domain gap between CLIP's pretraining space and point clouds, we propose cross-domain calibration (CDC), which efficiently bridges the manifold misalignment through source-domain semantic transfer and establishes a hybrid semantic space, enabling a joint embedding of 2D and 3D representations. Subsequently, ZUMA performs dynamic semantic interaction (DSI) to enable structural decoupling of anomaly regions in the high-dimensional embedding space constructed by CDC, where natural languages serve as semantic anchors to help DSI establish discriminative hyperplanes within hybrid modality representations. Within this framework, ZUMA enables plug-and-play detection of 2D, 3D or multimodal anomalies, without training or fine-tuning even for cross-dataset or incomplete-modality scenarios. Additionally, to further investigate the potential of the training-free ZUMA within the training-based paradigm, we develop ZUMA-FT, a fine-tuned variant that achieves notable improvements with minimal parameter trade-off. Extensive experiments are conducted on two MAD benchmarks, MVTec 3D-AD and Eyecandies. Notably, the training-free ZUMA achieves state-of-the-art (SOTA) performance on both datasets, outperforming existing ZS MAD methods, including training-based approaches. Moreover, ZUMA-FT further extends the performance boundary of ZUMA with only 6.75 M learnable parameters.

Person re-identification (re-id) models are vital in security surveillance systems, requiring transferable adversarial attacks to explore the vulnerabilities of them. Recently, vision-language models (VLM) based attacks have shown superior transferability by attacking generalized image and textual features of VLM, but they lack comprehensive feature disruption due to the overemphasis on discriminative semantics in integral representation. In this paper, we introduce the Attribute-aware Prompt Attack (AP-Attack), a novel method that leverages VLM's image-text alignment capability to explicitly disrupt fine-grained semantic features of pedestrian images by destroying attribute-specific textual embeddings. To obtain personalized textual descriptions for individual attributes, textual inversion networks are designed to map pedestrian images to pseudo tokens that represent semantic embeddings, trained in the contrastive learning manner with images and a predefined prompt template that explicitly describes the pedestrian attributes. Inverted benign and adversarial fine-grained textual semantics facilitate attacker in effectively conducting thorough disruptions, enhancing the transferability of adversarial examples. Extensive experiments show that AP-Attack achieves state-of-the-art transferability, significantly outperforming previous methods by 22.9% on mean Drop Rate in cross-model&dataset attack scenarios.

Monocular 3D visual grounding is a novel task that aims to locate 3D objects in RGB images using text descriptions with explicit geometry information. Despite the inclusion of geometry details in the text, we observe that the text embeddings are sensitive to the magnitude of numerical values but largely ignore the associated measurement units. For example, simply equidistant mapping the length with unit 'meters' to 'decimeters' or 'centimeters' leads to severe performance degradation, even though the physical length remains equivalent. This observation signifies the weak 3D comprehension of pre-trained language model, which generates misguiding text features to hinder 3D perception. Therefore, we propose to enhance the 3D perception of model on text embeddings and geometry features with two simple and effective methods. Firstly, we introduce a pre-processing method named 3D-text Enhancement (3DTE), which enhances the comprehension of mapping relationships between different units by augmenting the diversity of distance descriptors in text queries. Next, we propose a Text-Guided Geometry Enhancement (TGE) module to further enhance the 3D-text information by projecting the basic text features into geometrically consistent space. These 3D-enhanced text features are then leveraged to precisely guide the attention of geometry features. We evaluate the proposed method through extensive comparisons and ablation studies on the Mono3DRefer dataset. Experimental results demonstrate substantial improvements over previous methods, achieving new state-of-the-art results with a notable accuracy gain of 11.94% in the 'Far' scenario. Our code will be made publicly available.

Vision-language models like CLIP have revolutionized person re-identification (ReID) by enabling cross-modal semantic alignment. However, most of the existing CLIP-based ReID methods suffer from a critical limitation: semantic entanglement, where identity and attribute features are indiscriminately compressed into a single, undifferentiated token representation. This oversight fails to account for their inherently distinct roles in characterizing individuals.To address this limitation, we propose an Identity-Attribute-Decoupled Tokenization (IADT) method, a hierarchical framework with two synergistic components:Subject-oriented tokens that model identity through a cross-modality feature inverse mapping paradigm, preserving invariant biometric features;Attribute-aware tokens that capture localized characteristics through the cross-interaction of local features and learnable prototype vectors, dynamically focusing on discriminative regions without manual supervision.The hierarchical tokenization enables disentangled yet complementary representation learning: Identity and attribute semantics are encoded into distinct embedding subspaces, while cross-token contrastive learning establishes semantic reinforcement through attention-guided feature interaction. Crucially, this process does not require part-level annotations, making it directly applicable to real-world deployment. Extensive experiments validate effectiveness of the proposed method. For example, on the Market-1501 dataset, IADT achieves 97.1% mAP (+2.5% over SOTA) and 98.2% Rank-1 accuracy. For the challenging MSMT benchmark, it attains 88.9% mAP (+1.7% improvement) with 93.1% Rank-1 accuracy, demonstrating consistent superiority. The code will be available at https://github.com/llraay/IADT.

State-of-the-art defect segmentation methods rely on sufficient training data and struggle to generalize to unseen categories. Few-Shot Semantic Segmentation (FSS) is introduced to specifically address these issues. However, existing FSS models still face two challenges in the industry. 1) Defects usually present as weak features, resulting in incomplete segmentation; 2) Severe background interference often leads to incorrect segmentation. To tackle these problems, we propose the Multi-Context Aggregation Network (MCANet). Specifically, we design a Cross-Layer Multi-Level Feature Aggregation Module (CMAM). CMAM effectively aggregates discretely distributed multi-level defect features across different layers and guides the query image to perceive defects from the pixel level, which avoids incomplete segmentation caused by weak features. Additionally, a Foreground Correction Module (FCM) is developed, which is equipped with a dedicated background predictor (BP) and a foreground corrector (FC). BP places more emphasis on learning features from backgrounds rather than defects. FC achieves efficient feature ensemble and further suppresses the backgrounds misidentified as defects in CMAM. They collaborate to prevent incorrect segmentation caused by background interference. Extensive experiments demonstrate the effectiveness of our method. We achieve state-of-the-art results on both FSSD-12, a public benchmark FSS dataset for strip steel, and FSS-AEB, an FSS dataset for aero-engine blades. Specifically, with 1/5 support images, we achieve 64.6%/65.6% mIoU on FSSD-12 and 55.0%/57.8% mIoU on FSS-AEB. Note to Practitioners—Surface defect segmentation has always been a hot topic in the industry. However, existing methods rely on sufficient training data and struggle to generalize to unseen categories, which significantly hinders the automation of defect segmentation. To address this problem, we propose MCANet for automated few-shot defect segmentation. It achieves effective segmentation for surface defects with limited data, even for unseen categories. Furthermore, MCANet achieves state-of-the-art results on two datasets from real-world industrial scenarios and also delivers significant improvements over the widely concerned large vision models. Finally, we integrate MCANet into an automated surface defect inspection platform consisting of an imaging system and a high-performance computing server for real-world performance validation.

Deep learning based person re-identification (re-id) models have been widely employed in surveillance systems. Recent studies have demonstrated that black-box single-modality and cross-modality re-id models are vulnerable to adversarial examples (AEs), leaving the robustness of multi-modality re-id models unexplored. Due to the lack of knowledge about the specific type of model deployed in the target black-box surveillance system, we aim to generate modality unified AEs for omni-modality (single-, cross- and multi-modality) re-id models. Specifically, we propose a novel Modality Unified Attack method to train modality-specific adversarial generators to generate AEs that effectively attack different omni-modality models. A multi-modality model is adopted as the surrogate model, wherein the features of each modality are perturbed by metric disruption loss before fusion. To collapse the common features of omnimodality models, Cross Modality Simulated Disruption approach is introduced to mimic the cross-modality feature embeddings by intentionally feeding images to non-corresponding modality-specific subnetworks of the surrogate model. Moreover, Multi Modality Collaborative Disruption strategy is devised to facilitate the attacker to comprehensively corrupt the informative content of person images by leveraging a multi modality feature collaborative metric disruption loss. Extensive experiments show that our MUA method can effectively attack the omni-modality re-id models, achieving 55.9%, 24.4%, 49.0% and 62.7% mean mAP Drop Rate, respectively.

Deep learning-based person re-identification (re-id) models are widely employed in surveillance systems and inevitably inherit the vulnerability of deep networks to adversarial attacks. Existing attacks merely consider cross-dataset and cross-model transferability, ignoring the cross-test capability to perturb models trained in different domains. To powerfully examine the robustness of real-world re-id models, the Meta Transferable Generative Attack (MTGA) method is proposed, which adopts meta-learning optimization to promote the generative attacker producing highly transferable adversarial examples by learning comprehensively simulated transfer-based cross-model&dataset&test black-box meta attack tasks. Specifically, cross-model&dataset black-box attack tasks are first mimicked by selecting different re-id models and datasets for meta-train and meta-test attack processes. As different models may focus on different feature regions, the Perturbation Random Erasing module is further devised to prevent the attacker from learning to only corrupt model-specific features. To boost the attacker learning to possess cross-test transferability, the Normalization Mix strategy is introduced to imitate diverse feature embedding spaces by mixing multi-domain statistics of target models. Extensive experiments show the superiority of MTGA, especially in cross-model&dataset and cross-model&dataset&test attacks, our MTGA outperforms the SOTA methods by 20.0% and 11.3% on mean mAP drop rate, respectively. The source codes are available at https://github.com/yuanbianGit/MTGA.

Visible-infrared person re-identification (VI-ReID) seeks to identify and match individuals across visible and infrared ranges within intelligent monitoring environments. Most current approaches predominantly explore a two-stream network structure that extract global or rigidly split part features and introduce an extra modality for image compensation to guide networks reducing the huge differences between the two modalities. However, these methods are sensitive to misalignment caused by pose/viewpoint variations and additional noises produced by extra modality generating. Within the confines of this articles, we clearly consider addresses above issues and propose a Cross-modality Semantic Consistency Learning (CSCL) network to excavate the semantic consistent features in different modalities by utilizing human semantic information. Specifically, a Parsing-aligned Attention Module (PAM) is introduced to filter out the irrelevant noises with channel-wise attention and dynamically highlight the semantic-aware representations across modalities in different stages of the network. Then, a Semantic-guided Part Alignment Module (SPAM) is introduced, aimed at efficiently producing a collection of semantic-aligned fine-grained features. This is achieved by incorporating parsing loss and division loss constraints, ultimately enhancing the overall person representation. Finally, an Identity-aware Center Mining (ICM) loss is presented to reduce the distribution between modality centers within classes, thereby further alleviating intra-class modality discrepancies. Extensive experiments indicate that CSCL outperforms the state-of-the-art methods on the SYSU-MM01 and RegDB datasets. Notably, the Rank-1/mAP accuracy on the SYSU-MM01 dataset can achieve 75.72%/72.08%.

Supervised person re-identification (re-id) methods require expensive manual labeling costs. Although unsupervised re-id methods can reduce the requirement of the labeled datasets, the performance of these methods is lower than the supervised alternatives. Recently, some weakly supervised learning-based person re-id methods have been proposed, which is a balance between supervised and unsupervised learning. Nevertheless, most of these models require another auxiliary fully supervised datasets or ignore the interference of noisy tracklets. To address this problem, in this work, we formulate a weakly supervised tracklet association learning (WS-TAL) model only leveraging the video labels. Specifically, we first propose an intra-bag tracklet discrimination learning (ITDL) term. It can capture the associations between person identities and images by assigning pseudo labels to each person image in a bag. And then, the discriminative feature for each person is learned by utilizing the obtained associations after filtering the noisy tracklets. Based on that, a cross-bag tracklet association learning (CTAL) term is presented to explore the potential tracklet associations between bags by mining reliable positive tracklet pairs and hard negative pairs. Finally, these two complementary terms are jointly optimized to train our re-id model. Extensive experiments on the weakly labeled datasets demonstrate that WS-TAL achieves 88.1% and 90.3% rank-1 accuracy on the MARS and DukeMTMC-VideoReID datasets respectively. The performance of our model surpasses the state-of-the-art weakly supervised models by a large margin, even outperforms some fully supervised re-id models.

Occluded person re-identification (Re-ID) is a challenging task, as various object-to-person (OTP) and person-to-person (PTP) occlusion scenarios cause diverse occlusion interference and target person feature loss problems in person matching. Most existing methods, which utilize auxiliary models to evaluate the unoccluded person parts for occlusion feature elimination, are inefficient and cannot handle the PTP occlusion scenarios and person feature loss problems. To solve these issues, we propose a novel Occlusion-Aware Feature Recover (OAFR) model. OAFR simulates diverse occlusions to facilitate the model perceiving OTP, PTP occlusions and recovers occluded query features with unoccluded retrieved gallery features. Concretely, the Prior Knowledge-based Occlusion Simulation method is firstly introduced to synthesize OTP, PTP occlusions and corresponding occlusion labels, empowering model target person perception and occlusion-aware capability through self-supervised learning. Afterward, the feature recovery module reconstructs occluded query features with corresponding unoccluded local features of the top-$K$retrieved images by the visibility weighted average scheme, thus recovering the occluded query features to maintain more comprehensive features for better retrieval. Extensive experiments demonstrate that the proposed OAFR achieves superior performance to the state-of-the-art for both holistic and occluded Re-ID. Especially for Occluded-DukeMTMC dataset, OAFR outperforms the state-of-the-art by 6.0% for Rank-1 accuracy and 2.2% for mAP score. The source codes are available athttps://github.com/yuanbianGit/OAFR.