Query：
学者姓名：毋立芳
Refining：
Year
Type
Indexed by
Source
Complex
Co-Author
Language
Clean All
Abstract ：
Face Anti-Spoofing (FAS) plays a critical role in safeguarding face recognition systems, while previous FAS methods suffer from poor generalization when applied to unseen domains. Although recent methods have made progress via domain generalization technology, they are still sensitive to variations in face quality caused by task-irrelevant factors like camera and illumination. In this paper, we propose a novel Quality-Invariant Domain Generalization method (QIDG) with a teacher-student architecture, which aligns liveness features into a quality-invariant space to alleviate interference from task-irrelated factors. Specifically, QIDG utilizes the teacher model to produce face quality representations, which serve as the guidance for the student model to explore the quality-invariant space. To seek this space, the student model devises two novel modules, i.e., a dual adversarial learning module (DAL) and a quality feature assembly module (QFA). The former produces domain-invariant liveness features and task-irrelated quality features. While the latter assembles these two features from the same faces into complete quality representations, as well as assembles these two features from living faces in different domains. In this way, QIDG not only achieves the alignment of the domain-invariant liveness features to the quality-invariant space, but also promotes compactness of living faces from different domains in the feature space. Extensive cross-domain experiments demonstrate the superiority of our method on five public databases.
Keyword ：
Quality-invariant space Quality-invariant space Adversarial learning Adversarial learning Face anti-spoofing Face anti-spoofing Domain generalization Domain generalization
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Liu, Yongluo , Li, Zun , Xu, Yaowen et al. Quality-Invariant Domain Generalization for Face Anti-Spoofing [J]. | INTERNATIONAL JOURNAL OF COMPUTER VISION , 2024 , 132 (11) : 5239-5254 . |
| MLA | Liu, Yongluo et al. "Quality-Invariant Domain Generalization for Face Anti-Spoofing" . | INTERNATIONAL JOURNAL OF COMPUTER VISION 132 . 11 (2024) : 5239-5254 . |
| APA | Liu, Yongluo , Li, Zun , Xu, Yaowen , Guo, Zhizhi , Zou, Zhaofan , Wu, Lifang . Quality-Invariant Domain Generalization for Face Anti-Spoofing . | INTERNATIONAL JOURNAL OF COMPUTER VISION , 2024 , 132 (11) , 5239-5254 . |
| Export to | NoteExpress RIS BibTex |
Abstract ：
Image and text retrieval is a crucial topic in the fields of language and vision. The key to successful Image-Text retrieval is achieving accurate cross-modal representation and capturing essential correlations between image-sentence or words-regions. While existing work has designed intricate interactions to capture these correlations, challenges remain due to inadequate feature representations, such as insufficient text descriptions of image and ambiguous region representations. To address these challenges, we propose a novel approach, multi-view and region reasoning semantic enhancement, for image and text retrieval, which aims to enhance the semantic representation of features from both textual and visual modalities. Specifically, considering that an image can have multiple corresponding texts from different perspectives, with each text describing a single view, we devise a multi-view textual semantic enhancement module. This module takes advantage of the positive textual cues provided by corresponding image to make up for the limited knowledge in single-text views and produce a comprehensive image-based textual representation. Then, to address the semantic diversity of an image, we design a region reasoning semantic enhancement module that employs a graph structure to integrate both semantic and spatial reasoning knowledge from different regions, thereby clarifying the semantics of image regions and enhancing the overall semantic understanding of these areas. Extensive experiments and analyses demonstrate the superior performance of the proposed method on the Flickr30K and MSCOCO datasets, validating the effectiveness of the proposed solution.
Keyword ：
Semantic enhancement Semantic enhancement Image-text retrieval Image-text retrieval Graph Reasoning Graph Reasoning
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Cheng, Wengang , Han, Ziyi , He, Di et al. Multi-view and region reasoning semantic enhancement for image-text retrieval [J]. | MULTIMEDIA SYSTEMS , 2024 , 30 (4) . |
| MLA | Cheng, Wengang et al. "Multi-view and region reasoning semantic enhancement for image-text retrieval" . | MULTIMEDIA SYSTEMS 30 . 4 (2024) . |
| APA | Cheng, Wengang , Han, Ziyi , He, Di , Wu, Lifang . Multi-view and region reasoning semantic enhancement for image-text retrieval . | MULTIMEDIA SYSTEMS , 2024 , 30 (4) . |
| Export to | NoteExpress RIS BibTex |
Abstract ：
Digital light processing (DLP) 3D printing has attracted significant attention for its rapid printing speed, high accuracy, and diverse applications. However, the continuous DLP printing process releases substantial heat, resulting in a swift temperature rise in the curing area, which may lead to printing failures. Due to the lack of effective means to measure real-time temperature changes of the curing surface during continuous DLP 3D printing, the prevailing approach is to predict temperature variations during printing via numerical simulation. Nevertheless, temperature prediction methods relying solely on numerical simulation tend to be slow and overlook heat exchange dynamics during printing, potentially resulting in prediction inaccuracies, particularly for complex models. To address these issues, this paper proposes a method to combine numerical simulation and a machine learning approach for temperature prediction in the DLP 3D printing process, along with a printing control scheme generation method. Firstly, the ( m + n )th order autocatalytic kinetic model considering the light intensity and the Beer-Lambert law are employed to formulate the heat calculation equation for the photopolymer resin curing reaction. Subsequently, a heat exchange calculation equation is established based on Fourier heat conduction law and Newton's cooling equation. A numerical simulation model for temperature changes during the printing process is then developed by integrating the heat calculation equation, heat exchange calculation equation, and measurement data from Photo-DSC. Furthermore, a temperature measurement device for the printing process is designed to validate the accuracy of the numerical simulation. Following this, an improved Long Short-term Memory (LSTM) network is proposed, using temperature change data generated by the numerical simulation model to train the network for rapid (2 x 10 -4 s/layer) prediction of temperature changes during printing. Finally, aiming for the shortest printing time, an optimized control scheme planning algorithm and a target function are designed based on the model's temperature change data and the monomer's flash point to ensure the temperature remains below this threshold. This algorithm can automatically generate the optimal printing control scheme for any model. Experimental results demonstrate that the proposed temperature prediction method can predict temperature variation accurately. Based on this, the generated printing control scheme can guarantee efficient and high-quality manufacturing for any model.
Keyword ：
Temperature prediction Temperature prediction Numerical simulation Numerical simulation Machine learning Machine learning Digital light processing Digital light processing
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Zhao, Lidong , Zhao, Zhi , Ma, Limin et al. Developing the optimized control scheme for digital light processing 3D printing by combining numerical simulation and machine learning-guided temperature prediction [J]. | JOURNAL OF MANUFACTURING PROCESSES , 2024 , 132 : 363-374 . |
| MLA | Zhao, Lidong et al. "Developing the optimized control scheme for digital light processing 3D printing by combining numerical simulation and machine learning-guided temperature prediction" . | JOURNAL OF MANUFACTURING PROCESSES 132 (2024) : 363-374 . |
| APA | Zhao, Lidong , Zhao, Zhi , Ma, Limin , Li, Shuyi , Men, Zening , Wu, Lifang . Developing the optimized control scheme for digital light processing 3D printing by combining numerical simulation and machine learning-guided temperature prediction . | JOURNAL OF MANUFACTURING PROCESSES , 2024 , 132 , 363-374 . |
| Export to | NoteExpress RIS BibTex |
Abstract ：
Face anti-spoofing (FAS) plays a vital role in securing face recognition systems. Previous approaches usually learn spoofing features from a single perspective, in which only universal cues shared by all attack types are explored. However, such single-perspective-based approaches ignore the differences among various attacks and commonness between certain attacks and bona fides, thus tending to neglect some non-universal cues that contain strong discernibility against certain types. As a result, when dealing with multiple types of attacks, the above approaches may suffer from the uncomprehensive representation of bona fides and spoof faces. In this work, we propose a novel Advanced Multi-Perspective Feature Learning network (AMPFL), in which multiple perspectives are adopted to learn discriminative features, to improve the performance of FAS. Specifically, the proposed network first learns universal cues and several perspective-specific cues from multiple perspectives, then aggregates the above features and further enhances them to perform face anti-spoofing. In this way, AMPFL obtains features that are difficult to be captured by single-perspective-based methods and provides more comprehensive information on bona fides and spoof faces, thus achieving better performance for FAS. Experimental results show that our AMPFL achieves promising results in public databases, and it effectively solves the issues of single-perspective-based approaches.
Keyword ：
multi-perspective multi-perspective universal cues universal cues Face anti-spoofing Face anti-spoofing
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Wang, Zhuming , Xu, Yaowen , Wu, Lifang et al. Improving Face Anti-spoofing via Advanced Multi-perspective Feature Learning [J]. | ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS , 2023 , 19 (6) . |
| MLA | Wang, Zhuming et al. "Improving Face Anti-spoofing via Advanced Multi-perspective Feature Learning" . | ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS 19 . 6 (2023) . |
| APA | Wang, Zhuming , Xu, Yaowen , Wu, Lifang , Han, Hu , Ma, Yukun , Li, Zun . Improving Face Anti-spoofing via Advanced Multi-perspective Feature Learning . | ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS , 2023 , 19 (6) . |
| Export to | NoteExpress RIS BibTex |
Abstract ：
In recent years, a series of continuous fabrication technologies based on digital light processing (DLP) 3D printing have emerged, which have significantly improved the speed of 3D printing. However, limited by the resin filling speed, those technologies are only suitable to print hollow structures. In this paper, an optimized protocol for developing continuous and layer-wise hybrid DLP 3D printing mode is proposed based on computational fluid dynamics (CFD). Volume of the fluid method is used to simulate the behavior of resin flow while Poiseuille flow, Jacobs working curve, and Beer-Lambert law are used to optimize the key control parameters for continuous and layer-wise printing. This strategy provides a novel simulation-based method development scenario to establish printing control parameters that are applicable to arbitrary structures. Experiments verified that the printing control parameters obtained by simulations can effectively improve the printing efficiency and the applicability of DLP 3D printing.
Keyword ：
Computational fluid dynamics Computational fluid dynamics Resin filling Resin filling Control parameters Control parameters DLP 3D printing DLP 3D printing Continuous printing Continuous printing Layer-wise printing Layer-wise printing
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Zhao, Lidong , Zhang, Yan , Wu, Lifang et al. Developing the optimized control scheme for continuous and layer-wise DLP 3D printing by CFD simulation [J]. | INTERNATIONAL JOURNAL OF ADVANCED MANUFACTURING TECHNOLOGY , 2023 , 125 (3-4) : 1511-1529 . |
| MLA | Zhao, Lidong et al. "Developing the optimized control scheme for continuous and layer-wise DLP 3D printing by CFD simulation" . | INTERNATIONAL JOURNAL OF ADVANCED MANUFACTURING TECHNOLOGY 125 . 3-4 (2023) : 1511-1529 . |
| APA | Zhao, Lidong , Zhang, Yan , Wu, Lifang , Zhao, Zhi , Men, Zening , Yang, Feng . Developing the optimized control scheme for continuous and layer-wise DLP 3D printing by CFD simulation . | INTERNATIONAL JOURNAL OF ADVANCED MANUFACTURING TECHNOLOGY , 2023 , 125 (3-4) , 1511-1529 . |
| Export to | NoteExpress RIS BibTex |
Abstract ：
The growing complex user intention gap and information overload are obstacles for users to access the desired content. User interactions and the involved content indicate rich evidence of users' interests. It is required to investigate interaction characters over user interest and information distribution, and this alleviates information overload for personalized recommendation. Therefore, this work explores user interests with interactions and visual information from users' historical records for image recommendation. This paper introduces cross-modal manifold propagation (CMP) for personalized image recommendation. CMP investigates the trend of user preferences by propagating users' historical records along with users' interest distribution, which produces personalized interest-aware image candidates according to user interests. CMP simultaneously leverages visual distribution to spread users' visual records relying on the dense semantic visual manifold. Visual manifold propagation estimates detailed semantic-level user-image correlations for ranking candidate images in recommendations. In the proposed CMP, both user interest manifold and images' visual manifold compensate each other in propagating users' records to predict users' interaction. Experimental results illustrate the effectiveness of collaborative user-image propagation of CMP for personalized image recommendation. Performance improved by more than 20% compared to that of existing baselines.
Keyword ：
user interest user interest manifold propagation manifold propagation cross-modal collaboration cross-modal collaboration personalized recommendation personalized recommendation collaborative propagation collaborative propagation
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Jian, Meng , Guo, Jingjing , Fu, Xin et al. Cross-Modal Manifold Propagation for Image Recommendation [J]. | APPLIED SCIENCES-BASEL , 2022 , 12 (6) . |
| MLA | Jian, Meng et al. "Cross-Modal Manifold Propagation for Image Recommendation" . | APPLIED SCIENCES-BASEL 12 . 6 (2022) . |
| APA | Jian, Meng , Guo, Jingjing , Fu, Xin , Wu, Lifang , Jia, Ting . Cross-Modal Manifold Propagation for Image Recommendation . | APPLIED SCIENCES-BASEL , 2022 , 12 (6) . |
| Export to | NoteExpress RIS BibTex |
Abstract ：
In anchor-free object detection, the center regions of bounding boxes are often highly weighted to enhance detection quality. However, the central area may become less significant in some situations. In this paper, we propose a novel dual attention-based approach for the adaptive weight assignment within bounding boxes. The proposed improved dual attention mechanism allows us to thoroughly untie spatial and channel attention and resolve the confusion issue, thus it becomes easier to obtain the proper attention weights. Specifically, we build an end-to-end network consisting of backbone, feature pyramid, adaptive weight assignment based on dual attention, regression, and classification. In the adaptive weight assignment module based on dual attention, a parallel framework with the depthwise convolution for spatial attention and the 1D convolution for channel attention is applied. The depthwise convolution, instead of standard convolution, helps prevent the interference between spatial and channel attention. The 1D convolution, instead of fully connected layer, is experimentally proved to be both efficient and effective. With the adaptive and proper attention, the correctness of object detection can be further improved. On public MS-COCO dataset, our approach obtains an average precision of 52.7%, achieving a great increment compared with other anchor-free object detectors.
Keyword ：
dual attention dual attention anchor-free object detection anchor-free object detection adaptive weight assignment adaptive weight assignment
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Xiang, Ye , Zhao, Boxuan , Zhao, Kuan et al. Improved Dual Attention for Anchor-Free Object Detection [J]. | SENSORS , 2022 , 22 (13) . |
| MLA | Xiang, Ye et al. "Improved Dual Attention for Anchor-Free Object Detection" . | SENSORS 22 . 13 (2022) . |
| APA | Xiang, Ye , Zhao, Boxuan , Zhao, Kuan , Wu, Lifang , Wang, Xiangdong . Improved Dual Attention for Anchor-Free Object Detection . | SENSORS , 2022 , 22 (13) . |
| Export to | NoteExpress RIS BibTex |
Abstract ：
Hamming space retrieval is a hot area of research in deep hashing because it is effective for large-scale image retrieval. Existing hashing algorithms have not fully used the absolute boundary to discriminate the data inside and outside the Hamming ball, and the performance is not satisfying. In this paper, a boundary-aware contrastive loss is designed. It involves an exponential function with absolute boundary (i.e., Hamming radius) information for dissimilar pairs and a logarithmic function to encourage small distance for similar pairs. It achieves a push that is bigger than the pull inside the Hamming ball, and the pull is bigger than the push outside the ball. Furthermore, a novel Boundary-Aware Hashing (BAH) architecture is proposed. It discriminatively penalizes the dissimilar data inside and outside the Hamming ball. BAH enables the influence of extremely imbalanced data to be reduced without up-weight to similar pairs or other optimization strategies because its exponential function rapidly converges outside the absolute boundary, making a huge contrast difference between the gradients of the logarithmic and exponential functions. Extensive experiments conducted on four benchmark datasets show that the proposed BAH obtains higher performance for different code lengths, and it has the advantage of handling extremely imbalanced data.
Keyword ：
Hamming space retrieval Hamming space retrieval boundary-aware penalization boundary-aware penalization image retrieval image retrieval deep hashing deep hashing
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Hu, Wenjin , Chen, Yukun , Wu, Lifang et al. Boundary-Aware Hashing for Hamming Space Retrieval [J]. | APPLIED SCIENCES-BASEL , 2022 , 12 (1) . |
| MLA | Hu, Wenjin et al. "Boundary-Aware Hashing for Hamming Space Retrieval" . | APPLIED SCIENCES-BASEL 12 . 1 (2022) . |
| APA | Hu, Wenjin , Chen, Yukun , Wu, Lifang , Shi, Ge , Jian, Meng . Boundary-Aware Hashing for Hamming Space Retrieval . | APPLIED SCIENCES-BASEL , 2022 , 12 (1) . |
| Export to | NoteExpress RIS BibTex |
Abstract ：
Group activity recognition is a subject with broad applications, and its main challenge is to model the interactions between individuals. Existing algorithms mostly model the interactions merely based on holistic features of persons, which completely ignore the local details and local interactions that could be significant for recognition. In this paper, we propose a novel part based interaction learning algorithm for group activity recognition. Our proposed algorithm introduces both the physical structural information and fine-grained contextual information into representations, through exploring the intra-and inter-actor part interactions. Specifically, a dual-branch framework is adopted to extract the appearance and motion features respectively. For each branch, we utilize the key point detection technique for proper part division and then extract the part features. The part features are further enhanced by the transformers for intra-and inter-actor part interactions, and are lastly used for group activity recognition. Comparison with the state-of-the-arts on two public datasets demonstrate the effectiveness of our proposed algorithm.
Keyword ：
Group activity recognition Group activity recognition part based interaction part based interaction
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Tian, Meng , Lang, Xianglong , Xiang, Ye et al. Part Based Interaction Learning for Group Activity Recognition [J]. | 2022 IEEE 24TH INTERNATIONAL WORKSHOP ON MULTIMEDIA SIGNAL PROCESSING (MMSP) , 2022 . |
| MLA | Tian, Meng et al. "Part Based Interaction Learning for Group Activity Recognition" . | 2022 IEEE 24TH INTERNATIONAL WORKSHOP ON MULTIMEDIA SIGNAL PROCESSING (MMSP) (2022) . |
| APA | Tian, Meng , Lang, Xianglong , Xiang, Ye , Huang, Yan , Wu, Lifang , Li, Yunxiang . Part Based Interaction Learning for Group Activity Recognition . | 2022 IEEE 24TH INTERNATIONAL WORKSHOP ON MULTIMEDIA SIGNAL PROCESSING (MMSP) , 2022 . |
| Export to | NoteExpress RIS BibTex |
Abstract ：
Scale variation is one of the key challenges in the object detection. Most previous object detectors remedy this by using dilated convolution to enlarge the receptive fields of the vanilla convolutional layers. However, these methods focus on either the spatial information of small objects or the semantics of middle and large objects, which still fail to effectively adapt the scale variance of different objects, resulting in a sub-optimal performance for the object detection. In this paper, we propose a novel Multi-Sequence Dilated Network (MSDN) that stacks different dilated convolutions with different orders in parallel for improving the performance of the object detection. Concretely, MSDN contains a sequential dilated module and a dilated attention module. The former aims to generate scale-specific feature maps with fine-spatial and semantic information of objects at different scales, while the latter further selects more powerful information to adaptively enlarge the receptive fields of object features at different scales. Facilitated with these modules, MSDN well obtains the fine-spatial and semantic information of objects at different scales, thus solving the problem of the scale variation. Comprehensive experimental results over two public object detection benchmarks clearly demonstrate the effectiveness of our proposed MSDN. Particularly, on the COCO dataset, the mAP value of MSDN is 48.7%, outperforming existing state-of-the-art methods in a single model manner.
Keyword ：
dilated convolution dilated convolution object detection object detection scale variation scale variation FPN FPN
Cite：
Copy from the list or Export to your reference management。
| GB/T 7714 | Cui, Di , Li, Zun , Xin, Chang et al. Multi-Sequence Dilated Network for Object Detection [J]. | 2022 IEEE 24TH INTERNATIONAL WORKSHOP ON MULTIMEDIA SIGNAL PROCESSING (MMSP) , 2022 . |
| MLA | Cui, Di et al. "Multi-Sequence Dilated Network for Object Detection" . | 2022 IEEE 24TH INTERNATIONAL WORKSHOP ON MULTIMEDIA SIGNAL PROCESSING (MMSP) (2022) . |
| APA | Cui, Di , Li, Zun , Xin, Chang , Wu, Lifang , Liu, Yongluo . Multi-Sequence Dilated Network for Object Detection . | 2022 IEEE 24TH INTERNATIONAL WORKSHOP ON MULTIMEDIA SIGNAL PROCESSING (MMSP) , 2022 . |
| Export to | NoteExpress RIS BibTex |
Export
| Results： |
Selected to |
| Format： |
