Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks334 from public data
- Disentangling and Unifying Graph Convolutions for Skeleton-Based Action Recognition1,225
A simple method to disentangle multi-scale graph convolutions and a unified spatial-temporal graph convolutional operator named G3D are presented and a powerful feature extractor named MS-G3D is developed based on which the model outperforms previous state-of-the-art methods on three large-scale datasets.
- Study on subset size selection in digital image correlation for speckle patterns646
A theoretical model of the displacement measurement accuracy of DIC can be accurately predicted based on the variance of image noise and Sum of Square of Subset Intensity Gradients (SSSIG), which leads to a simple criterion for choosing an optimal subset size for the DIC analysis.
- Video summarization via minimum sparse reconstruction185
This paper formulate the video summarization task with a novel minimum sparse reconstruction (MSR) problem, where the original video sequence can be best reconstructed with as few selected keyframes as possible.
- 174
- 157
- Learning visual relationship and context-aware attention for image captioning154
A novel method to implicitly model the relationship among regions of interest in an image with a graph neural network, as well as a novel context-aware attention mechanism to guide attention selection by fully memorizing previously attended visual content are proposed.
- 149
- 136
- Graph weeds net: A graph-based deep learning method for weed recognition126
A novel graph-based deep learning architecture, namely Graph Weeds Net (GWN), which aims to recognize multiple types of weeds from conventional RGB images collected from complex rangelands, and provides suggestions for key regions, creating opportunities for further within-image actions for robotic in-field systems.
- 112
- Leaf Image Retrieval with Shape Features103
An efficient two-step approach of using a shape characterization function called centroid-contour distance curve and the object eccentricity (or elongation) for leaf image retrieval and Experimental results show that this approach can achieve good performance with a reasonable computational complexity.
- 99
- Deep learning techniques for in-crop weed recognition in large-scale grain production systems: a review93
This paper reviews recent developments of deep learning techniques in the field of image-based weed detection, and the challenges of developing practically deployable weed detection methods are summarized.
- 92
- The Codecfake Dataset and Countermeasures for the Universally Detection of Deepfake Audio90
This work demonstrates that ADD model training with the Codecfake dataset can effectively detects ALM-based audio, and proposes the CSAM strategy to learn a domain balanced and generalized minima to achieve universal detection of deepfake audio.
- 86
- 82
- Video summarization via block sparse dictionary selection78
Experimental results on two benchmark datasets demonstrate that the proposed SBOMP based VS method clearly outperforms several state-of-the-art sparse representation based methods in terms of F-score, redundancy among keyframes and robustness to outlier frames.
- 71
- Weed detection to weed recognition: reviewing 50 years of research to identify constraints and opportunities for large-scale cropping systems65
This review looks at the progress made over this half-century of research, summarizing advances in computer vision techniques and the most recent deep convolutional neural network approaches to weed recognition; and shows that the variability of large-scale cropping systems is best managed by deep learning for in-crop weed recognition.
- 65
- Sign Language Translation with Hierarchical Spatio-Temporal Graph Neural Network64
A novel deep learning architecture, namely hierarchical spatio-temporal graph neural network (HST-GNN), is proposed and graph convolutions and graph self-attentions with neighborhood context are proposed to characterize both the local and the global graph properties.
- Stacked Memory Network for Video Summarization63
A stacked memory network called SMN is proposed to explicitly model the long dependency among video frames so that redundancy could be minimized in the video summaries produced and is particularly good at capturing long temporal dependency among frames with few additional training parameters.
- LiDARCap: Long-range Markerless 3D Human Motion Capture with LiDAR Point Clouds62
Qualitative and qualitative experiments show that the LiDARHuman26M method outperforms the techniques based only on RGB images, and the experiments on the KITTI Dataset and the Waymo Open dataset show that it can be generalized to different LiDar sensor settings.
- 60
- VLIPP: Towards Physically Plausible Video Generation with Vision and Language Informed Physical Prior57
A novel two-stage image-to-video generation framework that explicitly incorporates physics with vision and language informed physical prior and can produce physically plausible motion is proposed.
- 56
- 53
- 50
- A Top-Down Approach for Video Summarization50
This work proposes a top-down approach consisting of scene identification and scene summarization, and develops a visual word-based approach to make the approach more computationally scalable.
- 49
- 47
- Learning universal multiview dictionary for human action recognition46
A novel discriminative dictionary learning framework is proposed by formulating a universal dictionary which consists of a shared sub-dictionary and a set of class-specific sub-Dictionaries and is able to achieve better performance than a number of state-of-the-art ones.
- Eye tracking data guided feature selection for image classification46
Comprehensive experimental results in two benchmark datasets demonstrate that eye tracking data are of great importance to improve the performance of feature selection for image classification.
- 46
- 45
- A study on multi-kernel intuitionistic fuzzy C-means clustering with multiple attributes40
This paper proposes to address the issue of IFCM with multi-kernel mapping, where different types of features are grouped and a composite kernel is constructed to map each attribute group into an individual kernel space and to linearly combine these kernels with optimimal weights.
- 40
- 38
- 38
- 37
- 37
- Learning realistic facial expressions from web images36
A search based framework to collect realistic facial expression images from the Web so as to further advance research on robust facial expression recognition and a novel facial expression feature based on the state-of-the-art Weber Local Descriptor and histogram contextualization is proposed.
- 36
- Skin Lesion Recognition with Class-Hierarchy Regularized Hyperbolic Embeddings34
A hyperbolic network is proposed to learn image embeddings and class prototypes jointly and the learned prototypes preserve the semantic class relations in the embedding space and can predict the label of an image by assigning its feature to the nearesthyperbolic class prototype.
- 34
- 33
- Single-channel EEG based insomnia detection with domain adaptation32
This is the first deep learning based domain adaptation model using single channel raw EEG signals to detect insomnia at subject level and demonstrates that the model generalizes well on two target domain datasets with different sampling rates.
- 31
- 31
- Adversarial Evolving Neural Network for Longitudinal Knee Osteoarthritis Prediction30
A novel deep learning architecture, namely adversarial evolving neural network (A-ENN), is proposed for longitudinal grading of KOA severity and can be applied to other learning tasks together with existing deep architectures, in which the responses characterize progressive representations.
- 30
- HGCLIP: Exploring Vision-Language Models with Graph Representations for Hierarchical Understanding29
A novel framework that effectively combines CLIP with a deeper exploitation of the Hierarchical class structure via Graph representation learning is proposed, which explores constructing the class hierarchy into a graph, with its nodes representing the textual or image features of each category.
- 28
- Hand vein recognition based on the connection lines of reference point and feature point28
Experimental results show that the proposed algorithm is able to achieve hand vein recognition reliably and quickly and effectively overcome the influence on the recognition results caused by image translation and rotation.
- Intelligent Grading of Tobacco Leaves Using an Improved Bilinear Convolutional Neural Network27
This work developed assembly line equipment and an RGB image classification method for tobacco grading that uses a pyramid structure, attention mechanism, and angle decision loss function to improve the bilinear convolutional neural network (a fine-grained visual categorization framework) for Tobacco grading.
- 27
- HDS-SP: A novel descriptor for skeleton-based human action recognition27
A HDS-SP descriptor for skeleton-based human action, i.e., the histogram of distributed sectors based on their specific projections, based on the principle of the better viewpoint the better action recognition performance is devises.
- Real-time hand posture recognition using hand geometric features and Fisher Vector27
The experimental results on four public HPR datasets show that the proposed method can achieve the mean accuracy of the state-of-the-art methods in real time.
- 26
- Continuous Intermediate Token Learning with Implicit Motion Manifold for Keyframe Based Motion Interpolation26
This paper proposes a novel framework to formulate latent motion manifolds with keyframe-based constraints, from which the continuous nature of intermediate token representations is considered, and demonstrates both superior interpolation accuracy and high visual similarity to ground truth motions.
- Keyframe Extraction from Motion Capture Sequences with Graph based Deep Reinforcement Learning26
This work proposes a novel graph based deep reinforcement learning method for efficient unsupervised keyframe selection that is able to select keyframes at a high efficiency without clearly compromising the quality in comparison with the state-of-the-art methods.
- Spectral embedding based facial expression recognition with multiple features26
A spectral embedding based multi-view dimension reduction method to fuse multiple features for facial expression recognition, which provides a strong support to the effectiveness of the proposed feature fusion framework on realistic facial expressions.
- 24
- 24
- Detect All-Type Deepfake Audio: Wavelet Prompt Tuning for Enhanced Auditory Perception22
This paper is the first to comprehensively establish an all-type ADD benchmark to evaluate current CMs, incorporating cross-type deepfake detection across speech, sound, singing voice, and music and introduces the prompt tuning self-supervised learning (PT-SSL) training paradigm, which optimizes SSL front-end by learning specialized prompt tokens for ADD, requiring 458× fewer trainable parameters than fine-tuning (FT).
- Video Summarization via Nonlinear Sparse Dictionary Selection22
A nonlinear SDS model is formulated for VS, in which the nonlinearity is transformed to linearity by projecting a video to a high-dimensional feature space induced by a kernel function.
- 20
- Fine-grained Action Recognition with Robust Motion Representation Decoupling and Concentration20
A motion representation decoupling and concentration network (MDCNet) to address two key issues: decouple robust motion representations from the spatio-temporal features and further effectively leverage them to enhance the learning of discriminative features still remains less explored.
- 20
- 20
- 20
- 20
- Codecfake: An Initial Dataset for Detecting LLM-based Deepfake Audio19
To effectively detect LLM-based deepfake audio, this work proposes Codecfake dataset, which is generated by seven representative neural codec methods, and shows that codec-trained ADD models exhibit a 41.406% reduction in average equal error rate compared to vocoder-trained ADD models on the Codecfake test set.
- 19
- 18
- TLDW: Extreme Multimodal Summarization of News Videos18
A novel unsupervised Hierarchical Optimal Transport Network (HOT-Net) consisting of three components: hierarchical multimodal encoder, hierarchical multimodal fusion decoder, and optimal transport solver is proposed, which achieves promising performance in terms of ROUGE and IoU metrics.
- 18
- Investigating the impact of frame rate towards robust human action recognition18
Promising results indicate that well designed key-frame selection methods can produce a set of representative frames and eventually reduce the impact of frame rate on the performance of human action recognition.
- 18
- 17
- Deep3D reconstruction: methods, data, and challenges17
This work presents a comprehensive survey of 3D reconstruction methods based on deep learning, dividing them into four types, recurrent neural network, deep autoencoder, generative adversarial network, and convolutional neural network based methods, and investigates four representative databases that are commonly used by the above methods in detail.
- 17
- 17
- Automated Nuclear Segmentation of Coherent Anti-Stokes Raman Scattering Microscopy Images by Coupling Superpixel Context Information with Artificial Neural Networks17
A nuclear segmentation technique which is essential for the automated analysis of CARS images in differential diagnosis of lung cancer subtypes is developed by coupling superpixel context information and an artificial neural network (ANN), which is, to the best of the authors' knowledge, the first automated nuclear segmentations approach for CARS pictures.
- 16
- Region-based supervised annotation for semantic image retrieval16
This work forms image annotation as a multi-class classification problem under the multi-instance learning framework, which deals with the weak annotation problem and works with image-level ground truth training data.
- 16
- 16
- 15
- Research on EEG emotion recognition based on CNN+BiLSTM+self-attention model15
The model of convolutional neural network and bi-directional long short-term memory and self-attention (CNN+BiLSTM+self-att attention) is proposed to extract more distinctive features from both spatial and temporal dimensions.
- Vision-Based Freezing of Gait Detection with Anatomic Patch Based Representation15
This paper treats video based FoG detection as a fine grained human action recognition problem and reduces the interference of visual content which is irrelevant to FoG, such as gait motion of supporting staff involved in clinical assessment.
- Extraction of interference phase in frequency-scanning interferometry based on empirical mode decomposition and Hilbert transform15
A phase-extraction algorithm based on empirical mode decomposition and Hilbert transformation that can effectively eliminate the influence of the optical frequency scanning nonlinearity is proposed.
- Browse-to-Search15
This work develops a novel system to explore an effective and efficient way for allowing users to seamlessly switch between browse and search processes, and naturally complete visual-based exploratory search tasks, and provides users with more robust results to satisfy their exploring experience.
- 15
- Coupling matrix manifolds assisted optimization for optimal transport problems14
This paper introduces a new manifold named as the coupling matrix manifold (CMM), where each point on this novel manifold can be regarded as a transportation plan of the optimal transport problem.
- 14
- Machine-learning prediction of a novel diagnostic model using mitochondria-related genes for patients with bladder cancer13
A novel diagnostic model based on mitochondria-related genes (MRGs) for BC patients using Machine Learning was developed whose diagnostic value was confirmed in GSE13507, GSE3167 and GSE37816 datasets.
- 13
- 13
- InvolutionGAN: lightweight GAN with involution for unsupervised image-to-image translation13
A novel loss term is developed that evaluates the learned perceptual similarity distance to regulate the training process and could produce competitive image results while saving computational costs up to 91.9%.
- Multi-Level Adversarial Spatio-Temporal Learning for Footstep Pressure Based FoG Detection13
One of the first studies of footstep pressure-based FoG detection and the approach of utilizing ASTN is the first deep neural network architecture in pursuit of subject-independent representations, which helps to reduce the over-fitting risk due to the high inter-subject variance.
- 13
- Semantic context based refinement for news video annotation13
This paper presents a novel video annotation refinement approach by utilizing extrinsic semantic context extracted from video subtitles and intrinsic context among candidate annotation concepts and Random Walk with Restarts technique.
- Design your own universe: a physics-informed agnostic method for enhancing graph neural networks12
Empirical validations on benchmarks for homophilic, heterophilic graphs, and long-term graph datasets show that GNNs enhanced by the proposed model-agnostic enhancement framework significantly outperform their original counterparts.
- 12
- Semantic-enhanced graph neural network for named entity recognition in ancient Chinese books12
This paper proposes a novel model based on Graph Neural Networks that integrates two different forms of external knowledge: dictionary-level and chapter-level information that is effectively incorporated into the model’s input context.
- 12
- SurgicalPart-SAM: Part-to-Whole Collaborative Prompting for Surgical Instrument Segmentation12
This paper investigates text promptable surgical instrument segmentation and proposes SP-SAM, a novel efficient-tuning approach that integrates surgical instrument structure knowledge with the generic segmentation knowledge of SAM.
- 12
- 3D human pose estimation from range images with depth difference and geodesic distance12
This work proposes an effective and efficient framework for pose estimation from range images that outperforms several state-of-the-art pose estimation methods based on the geodesic distance of pairwise pixels.
- SCUT-MMSIG: A Multimodal Online Signature Database12
A multimodal online signature database (SCUT-MMSIG) is presented, consisting of three subcorpora, namely, mobile, tablet and in-air signature subcorpus, with 20 genuine samples and 20 skilled forgeries per subject.
- 12
- Adaptive Processing of Tree-Structure Image Representation12
In this paper, a segmentation-free tree-structure image representation is presented and a back-propagation through structure (BPTS) algorithm is adopted in order to learn the structure representation.
- 11
- 11
- Establishment of a Knowledge‐and‐Data‐Driven Artificial Intelligence System with Robustness and Interpretability in Laboratory Medicine11
An advanced intelligent system that enables a comprehensive interpretation of lab big data, which substantially improves the clinical diagnosis and exhibited outstanding transparency and interpretability in three universal clinical application scenarios and outperformed human physicians in interpreting lab reports.
- 11
- 11
- Three-dimensional cartoon facial animation based on art rules11
A 3D cartoon facial animation system is developed for a subject given only a single frontal face image of a neutral expression and it is proposed to utilize the tables of existing models through model mapping to avoid time-consuming construction of a face animation table.
- DSMNet: Deep High-Precision 3-D Surface Modeling From Sparse Point Cloud Frames10
A novel learning-based joint framework, DSMNet, is proposed for high-precision 3-D surface modeling from sparse point cloud frames that outperforms the state-of-the-art methods in PCS and PCR on the Multi-View Partial Point Cloud (MVP) database.
- 10
- Multimedia for Future Health—Smart Medical Home10
This chapter presents the latest research and development in multimedia technologies and the transition of these technologies into health care products for the smart medical home and applications involving multimedia technologies in biomedicine.
- 10
- 9
- 9
- 9
- Robust video summarization using collaborative representation of adjacent frames9
Experimental results on two benchmark datasets demonstrate that the proposed method not only outperforms the state of the art, but also improves the robustness to transitional frames and “outlier” frames.
- Exploiting spatial-temporal context for trajectory based action video retrieval9
This paper develops a descriptor coding method based on the correlation between spatial-temporal and feature aspects of individual trajectories among trajectories, and develops an offset-aware distance measure for improved trajectory matching.
- 9
- Discriminative two-level feature selection for realistic human action recognition9
Detailed experimental results on the widely used and challenging Hollywood-2 Human Action dataset and YouTube dataset demonstrate that the proposed method is computationally efficient while achieving improved performance in recognizing realistic human actions.
- CONTENT-BASED IMAGE RETRIEVAL WITH RELEVANCE FEEDBACK USING ADAPTIVE PROCESSING OF TREE-STRUCTURE IMAGE REPRESENTATION9
This paper proposes a novel content-based image retrieval approach with relevance feedback using adaptive processing of tree-structure image representation that integrates image representation and similarity measure in a single framework.
- 8
- Motion Keyframe Interpolation for Any Human Skeleton via Temporally Consistent Point Cloud Sampling and Reconstruction8
Point Cloud-based Motion Representation Learning (PC-MRL), an unsupervised approach to enabling cross-compatibility between skeletons for motion interpolation learning, and devise a temporal point-wise K-nearest neighbors loss for unsupervised learning.
- Exploring Coarse-to-Fine Action Token Localization and Interaction for Fine-grained Video Action Recognition8
A coarse-to-fine action token localization and interaction network, namely C2F-ALIN, that dynamically localizes the most informative tokens at a coarse granularity and then partitions these located tokens to a fine granularity for sufficient fine-grained spatio-temporal interaction.
- Robust Knowledge Adaptation for Federated Unsupervised Person ReID8
A federated unsupervised cluster-contrastive (FedUCC) learning method is proposed for Person ReID that enables the sharing of mutual knowledge among clients while retaining local domain-specific knowledge based on the kinds of network layers and their parameters.
- Multi-scale Features Fusion for the Detection of Tiny Bleeding in Wireless Capsule Endoscopy Images8
The proposed model has been evaluated on a clinical dataset that contains 20,000 wireless capsule endoscopy images with clinical annotation and has the highest AP and the best medical diagnosis performance compared to state-of-the-art multi-scale models.
- Feature covariance matrix-based dynamic hand gesture recognition8
A novel method using the feature covariance matrix for effective and efficient dynamic HGR by tracking hand keypoints in three successive frames and calculating their motion features efficiently, which also provides a compact representation of a dynamic hand gesture.
- Video Summarization with Visual and Semantic Features8
This paper presents a method combining both visual and semantic features which assign semantic concepts to video segments through automatic video annotation, and demonstrates that semantic coherence is very helpful for video summarization when being fused with different visual features.
- 8
- 8
- Discriminative Segment Focus Network for Fine-grained Video Action Recognition7
A Discriminative Segment Focus Network (DSFNet) is devised to mine the discriminability of segment correlations and localize discriminative action-relevant segments for fine-grained video action recognition.
- TopicCAT: Unsupervised Topic-Guided Co-Attention Transformer for Extreme Multimodal Summarisation7
This study proposes a novel Unsupervised Topic-guided Co-Attention Transformer (TopicCAT) neural network to produce extreme multimodal summaries for video-document pairs, and proposes an optimal transport-based optimisation scheme to evaluate summary coverage from a semantic distribution perspective at the topic-level.
- 7
- 7
- ERINet: Enhanced rotation-invariant network for point cloud classification7
The experimental results on rotation-augmented datasets demonstrate that the proposed ERINet outperforms other state-of-the-art methods in rotation robustness for point cloud classification.
- P-Norm Attention Deep CORAL: Extending Correlation Alignment Using Attention and the P-Norm Loss Function7
Attention to improve the quality of feature maps and the p-norm loss function to align the distribution of the source and target features, further reducing the offset caused by the classification loss function are proposed.
- DIC/Moiré hybrid method based on regular patterns for deformation measurement7
In this method, a Moiré fringe technique based on correlation coefficient is developed to provide accurate initial deformation estimation for DIC, indicating a higher computational efficiency and higher calculation accuracy than the conventional DIC method.
- 7
- 7
- Exploring the influence of feature representation for dictionary selection based video summarization7
- 7
- 7
- Fast human action classification and VOI localization with enhanced sparse coding7
Experimental results on both two popular benchmark datasets and the widely used localization dataset (MSR) demonstrate that the proposed approach reduces computational cost significantly while maintaining comparable classification accuracy to that of the state-of-the-art methods.
- 7
- Channel-shuffled transformers for cross-modality person re-identification in video6
A novel Channel-Shuffled Temporal Transformer (CSTT) is proposed for processing multi-frame sequences in conjunction with a ResNet backbone to form Hybrid Channel-Shuffled Transformer Net (HCSTNET), and the optimal partitioning of feature maps therein is investigated.
- Player-Team Heterogeneous Interaction Graph Transformer for Soccer Outcome Prediction6
HIGFormer (Heterogeneous Interaction Graph Transformer), a novel graph-augmented transformer-based deep learning model for soccer outcome prediction, is proposed, offering a new perspective on talent scouting and team strategy analysis.
- 6
- 6
- 6
- STA-Net: A Spatial–Temporal Joint Attention Network for Driver Maneuver Recognition, Based on In-Cabin and Driving Scene Monitoring6
A Spatial–Temporal Joint Attention Network (STA-Net) is proposed to realize high-efficient temporal and spatial feature extractions in driver maneuver recognition and introduces an asymmetric loss function to effectively tackle the issue of sample imbalance across diverse categories of driving maneuvers.
- 6
- 6
- 6
- 6
- Evaluating the impact of frame rate on video based human action recognition6
It is observed in this paper that frame rate does affect recognition performance and the impact of frame rate is not consistent across different scenarios, so better recognition approaches are demanded to further advance human action recognition towards more practical deployment.
- 6
- 6
- SHAPER: A General Architecture for Privacy-Preserving Primitives in Secure Machine Learning5
This paper proposes a hybrid Secret-sharing and Homomorphic encryption Architecture for Privacy-pERsevering machine learning (SHAPER), which protects sensitive data in encrypted or randomly shared domains instead of relying on a trusted third party.
- ProtoSimi: label correction for fine-grained visual categorization5
This paper proposes a novel learning framework named ProtoSimi, which employs an adaptive label correction strategy, ensuring effective learning on limited data and outperforms the existing label noisy methods by a large margin.
- 5
- 5
- Towards Efficient Visual Simplification of Computational Graphs in Deep Neural Networks5
This work proposes leveraging a suite of visual simplification techniques, including a cycle-removing method, a module-based edge-pruning algorithm, and an isomorphic subgraph stacking strategy that enhances the performance for recognizing and diagnosing DNN models.
- 5
- Visual tracking utilizing robust complementary learner and adaptive refiner5
This paper proposes a visual tracking algorithm by utilizing robust Complementary Learner and adaptive Refiner (CLR) and demonstrates that the proposed tracker outperforms other state-of-the-art methods with real-time speed.
- 5
- Unsupervised text segmentation using LDA and MCMC5
A data driven approach to text segmentation, while most of the existing unsupervised methods determine segmentation boundaries by empirically exploring similarity measurement between adjacent units by using Markov Chain Monte Carlo technique, is proposed.
- 5
- 5
- Uncertainty-guided attention learning for malaria parasite detection in thick blood smears4
An uncertainty-guided attention learning based network for malaria parasite detection from thick blood smears is proposed, which incorporates pixel attention mechanism to identify more fine-grained and pixel-wise informative features, to improve the classification capability of the model.
- 4
- Extended Short- and Long-Range Mesh Learning for Fast and Generalised Garment Simulation4
A novel GNN-based mesh learning framework with two key components to extend the message-passing range with minimal overhead, namely the Laplacian-Smoothed Dual Message-Passing (LSDMP) and the Geodesic Self-Attention (GSA) modules.
- Frameless Graph Knowledge Distillation4
This work provides a KD-based framework on multiscaled GNNs, known as graph framelet, and proves that by adequately utilizing the graph knowledge in a multiscaled manner provided by graph framelet decomposition, the student model is capable of adapting both homophilic and heterophilic graphs and has the potential of alleviating the oversquashing issue.
- 4
- Point Cloud Normal Estimation via Representation Learning on Height Maps4
A novel normal estimation method that first distinguishes between sharp and non-sharp feature points, creating a dual-branch normal estimation architecture that exploits noise-corrupted augmentations of 2D input height maps derived from a specific underlying surface.
- Unleash Graph Neural Networks from Heavy Tuning4
This work proposes a graph conditional latent diffusion framework (GNN-Diff) to generate high-performing GNNs directly by learning from checkpoints saved during a light-tuning coarse search, which produces GNN parameters that outperform those obtained through comprehensive grid search.
- When Graph Neural Networks Meet Dynamic Mode Decomposition4
This work illustrates how DMD can estimate a low-rank, finite-dimensional linear operator based on multiple states of the system, effectively approximating potential nonlinear interactions between nodes in the graph and introduces a family of DMD-GNN models that effectively leverage the low-rank eigenfunctions provided by the DMD algorithm.
- 4
- A Supervised Reinforcement Learning Algorithm for Controlling Drone Hovering4
A watcher-actor-critic (WAC) algorithm is applied to the drone’s hover control, which can quickly lock the exploration direction and achieve high robustness of the drone’s hover control while improving learning efficiency and reducing learning costs.
- 4
- 4
- 4
- 4
- 4
- 4
- Regularized Fuzzy Discriminant Analysis for Hyperspectral Image Classification With Noisy Labels4
A Regularized Fuzzy Discriminant Analysis (RFDA) based feature extraction method to effectively utilize the spatial and spectral information of HSIs with noisy labels and preserves the neighborhood relationship among the pixels in the spatial domain and the fundamental structure among the samples in the spectral-domain, which is beneficial for hyperspectral image classification.
- 4
- 4
- Method of Deep Web Collection for Mobile Application Store Based on Category Keyword Searching4
This paper proposes a collection method based on category keywords query to improve the crawl rate and integrity of the mobile app stores information collection and shows that this collection method effectively promoted information integrity and acquisition efficiency.
- 4
- 4
- 4
- An efficient retinex-like brightness normalization method for coding camera flashes and strong brightness variation in videos4
This paper proposes an efficient algorithm based on the retinex-like system which allows inter-frame brightness being normalized before applying the conventional motion estimation and compensation processes, and shows that this approach is superior to all similar approaches in the literature.
- 4
- 4
- 3
- 3
- Underwater Image Enhancement via Domain Adaptive Transfer Learning and Hybrid Reinforcement Model3
A novel underwater image restoration model via domain adaptive transfer learning and hybrid enhancement is proposed which integrates deep neural networks and the physical model into a whole framework based on the real blurred underwater images, clear on-land images, and the generated underwater images.
- SITransformer: Shared Information-Guided Transformer for Extreme Multimodal Summarization3
SITransformer is proposed, a Shared Information-guided Transformer for extreme multimodal summarization which involves a cross-modal shared information extractor and a cross-modal interaction module and significantly enhances the summarization quality for both video and text summaries for XMSMO.
- Siamese Biattention Pooling Network for Change Detection in Remote Sensing3
A novel deep learning architecture, namely Siamese Bi-Attention Pooling Network (SBA-PN), is proposed to emphasize broad-scale change patterns by exploiting both intraimage and interimage contexts and demonstrates the effectiveness with the state-of-the-art performance on two commonly used benchmark datasets.
- T2QRM: Text-Driven Quadruped Robot Motion Generation3
A text-driven quadrupedal robot motion generation framework (T2QRM) to generate realistic and diverse animal-like motions in simulation and introduces a novel FrameEncoder to overcome the challenge of generating appropriate and variable-length sequences.
- S2AC: Self-Supervised Attention Correlation Alignment Based on Mahalanobis Distance for Image Recognition3
S2AC is a novel approach for unsupervised deep domain adaptation that makes use of a stacked attention architecture as a feature map extractor, and is the first attempt at incorporating attention-based networks and self-supervised learning for image domain adaptation.
- 3
- 3
- 3
- Federated Visualization: A Privacy-preserving Strategy for Decentralized Visualization3
A novel privacy preservation strategy for decentralized visualization is presented, to imitate the flowchart of the federated learning framework, and reformulate the visualization process within a federated infrastructure.
- Speaker-Aware Monaural Speech Separation3
A novel speaker-aware monaural speech separation model that outperforms existing models in different separation scenarios is proposed and Experimental results on two WSJ0 mixed datasets demonstrate that the proposed model outperforms existing models in different separation scenarios.
- Path planning of UAV in mountain area’s forest rescuing3
A fuzzy clustering -PSO - genetic algorithm is presented and a method combined of both the shortest distance and point clustering is adopted, and it can resolve the path planning problem of the UAV for water environmental monitoring effectively.
- 3
- Dorsal hand vein recognition based on Gabor multi-orientation fusion and multi-scale HOG features3
An algorithm combining Gabor multi-orientation features fusion with Multi-scale Histogram of Oriented Gradient (MS-HOG) is proposed in this paper to improve the recognition accuracy and has good robustness in dorsal hand vein recognition.
- Discovering Commonness and Specificness for Human Action Recognition3
This paper proposes a novel discriminative dictionary learning method which recognizes the commonness and specificness among different action classes, and aims to obtain a universal dictionary which consists of two parts, a shared dictionary for all action classes and a set of class-specific dictionaries.
- What is happening3
This paper proposes a data-driven approach to verb oriented image annotation and re-rank the candidate verbs with the tag context discovered from the images which are both semantically and visually similar to the given image in the MIRFlickr dataset.
- 3
- 3
- 3
- SRENet: Spectral Re-Entry Network for Point Cloud Action Recognition2
SRENet is proposed, a spectral-aware framework designed to explicitly learn both global context and fine-grained temporal dynamics of motion from a frequency perspective for action recognition, validating the effectiveness of frequency modeling in point cloud-based action understanding.
- RPRA-ADD: Forgery Trace Enhancement-Driven Audio Deepfake Detection2
The RPRA-ADD is proposed, an integrated Reconstruction-Perception-Reinforcement-Attention networks based forgery trace enhancement-driven robust audio deepfake detection framework that surpasses existing methods in rigorous 3×3 cross-domain evaluations across Speech, Sound, and Singing, underscoring strong generalization capability across diverse audio domains.
- 2
- Digging into Depth and Color Spaces: A Mapping Constraint Network for Depth Super-Resolution2
A novel mapping constraint network (MCNet) that incorporates additional constraints derived from both LR depth maps and color images is introduced that aims to optimize the space of mapping functions and enhance the performance of DSR.
- 2
- 2
- 2
- Intelligent Fish Detection System with Similarity-Aware Transformer2
A new type of lightweight and plug-and-play edge intelligent vision system to automatically conduct fast fish detection with high-speed camera and a novel similarity-aware vision Transformer for fast fish detection (FishViT) is proposed to onboard identify every single fish in a dense and similar group.
- Diffusing to the Top: Boost Graph Neural Networks with Minimal Hyperparameter Tuning2
This work introduces a graph-conditioned latent diffusion framework (GNN-Diff) to generate high-performing GNNs based on the model checkpoints of sub-optimal hyperparameters selected by a light-tuning coarse search.
- A Feature Enhanced Scale-adaptive Convolutional Network for Ship Detection in Maritime Surveillance*2
- 2
- 2
- 2
- 2
- [Continuous contrast-enhanced ultrasound applied to acute kidney injury caused by sepsis: a diagnostic clinical study].2
Compared to SCr and BUN, contrast-enhanced ultrasound parameters can early response to renal dysfunction, and contribute to early diagnosis of sepsis induced AKI.
- 2
- 2
- Sparse Appearance Learning Based Automatic Coronary Sinus Segmentation in CTA2
This work proposes a multiscale sparse appearance learning based method for estimating vesselness towards automatically extracting the centerlines, which leverages the flexibility of sparse representation to model the inherent spatial coherence of vessel/background appearance and derive a vesselness measurement.
- StoryImaging2
The StoryImaging system is developed to illustrate a textual story with both images harvested from the Web and synthesized speech to improve information comprehension and provide engaging and explorative experiences for users by closely bridging textual and visual modalities.
- Discovering Semantics from Visual Information2
Various aspects of visual information annotation are discussed, including content representation, machine learning based annotation methodologies, and widely used datasets, and a novel data driven based approach is introduced to discover semantic relevance among concepts in visual domain.
- 2
- 2
- 2
- 2
- Quality of service routing strategy using supervised genetic algorithm2
A supervised genetic algorithm (SGA) is proposed to solve the quality of service (QoS) routing problems in computer networks and shows that SGA improves the ability of searching an optimum solution and accelerates the convergent process up to 20 times.
- 1
- DPI-TTS: Directional Patch Interaction for Fast-Converging and Style Temporal Modeling in Text-to-Speech1
Directional Patch Interaction for Text-to-Speech (DPI-TTS) is proposed, which builds on DiT and achieves fast training without compromising accuracy, and introduces a fine-grained style temporal modeling method that further improves speaker style similarity.
- 1
- 1
- 1
- 1
- 1
- 1
- XAI for In-hospital Mortality Prediction via Multimodal ICU Data1
An eXplainable Multimodal Mortality Predictor (X-MMP) approaching an efficient, explainable AI solution for predicting in-hospital mortality via multimodal ICU data, and introduces an explainable method, namely Layer-Wise Propagation to Transformer as a proper extension of the LRP method to Transformers.
- 1
- 1
- 1
- 1
- FCP Filter: A Dynamic Clustering-Prediction Framework for Customer Behavior1
FCP filter is proposed, a clustering-prediction framework based on FCP, which can forecast purchase behavior and filter random noise of individual transaction data and is superior in capturing temporal dynamics and having flexible number of groups.
- Lightweight optimization design of arm frame for aerial work platform based on NSGA-II algorithm1
A lightweight design that minimizes the variation stiffness while reducing weight is proposed and the weight of the optimized arm is reduced by the original 13.3%, lightweight effect is remarkable.
- Leukocyte recognition algorithm in leucorrhea microscopic images based on ResNet-34 neural network1
An automatic recognition algorithm based on ResNet-34 neural network is proposed to solve the problem of low efficiency of leucocyte recognition in traditional artificial microscopy.
- 1
- Learning Shared and Cluster-Specific Dictionaries for Single Image Super-Resolution1
This paper proposes a novel shared and cluster-specific dictionaries learning method for SISR, where cluster- specific dictionaries can be learned as separate as possible and a dictionary shared among clusters is learned explicitly to explore similarity among clusters.
- 1
- 1
- GW29-e2015 Primary endpoint Outcomes of A Prospective, Single-Blinded, Multi-Center, Randomized, Controlled, Clinical Trial on the Efficacy and Safety of Sirolimus-Eluting Stent (MiStent® System) compared to TIVOLI in the Treatment of Patients with Coronary Heart Disease1
Micell Technologies has developed the MiStent Sirolimus Eluting Absorbable Polymer Coronary Stent System (“MiStent”) consisting of a cobalt-chromium stent platform.
- 1
- 1
- 1
- 1
- 1
- –
- ProtoMix: Unified Prototype Learning for Semantic Change Detection in Remote Sensing–
This work proposes a novel prototype learning framework for SCD, namely ProtoMix, which functions through a Siamese encoder–decoder architecture with multitask learning for both change detection and semantic segmentation and accommodates high intraclass variance.
- –
- DSACap: Enhancing Visual-Semantic Alignment with Diffusion-based Framework for Image Captioning–
This work proposes a novel diffusion-based semantics aligned image captioning framework, namely DSACap, which deviates from existing methods which treat text as the target of noise-adding and denoising, instead directly applying these processes to the image, thus reducing the loss of visual-semantic alignment.
- –
- –
- –
- –
- –
- Towards an Effective Action-Region Tracking Framework for Fine-grained Video Action Recognition–
This work proposes a region-specific semantic activation module that employs discriminative and text-constrained semantics as queries to capture the most action-related region responses in each video frame, facilitating interaction among spatial and temporal dimensions with corresponding video features.
- –
- –
- Fast Online Adaptation of Visual SLAM via Variational Information Transfer and Preservation–
This work presents a novel variational information transfer and preservation based visual SLAM method that aims to adapt fast while maintaining good precision and surpasses several state-of-the-art baselines in evaluations of online visual SLAM adaptation.
- –
- Sight View Constraint for Robust Point Cloud Registration–
This paper proposes a novel and general Sight View Constraint (SVC) to conclusively identify incorrect transformations, thereby enhancing the robustness of existing PCR methods and highlighting the significance of the decision version problem of partial PCR.
- –
- A multi-speaker multi-lingual voice cloning system based on vits2 for limmits 2024 challenge–
The development of a speech synthesis system for the LIMMITS'24 Challenge, focusing primarily on Track 2, uses the VITS2 architecture, augmented with a multi-lingual ID and a BERT model to enhance contextual language comprehension.
- –
- –
- Constructing a Theory of Affective Stylistics in Modern Chinese Literature Based on Knowledge Mapping–
The emotion dictionary based on the point mutual information in the SOPMI algorithm is extended and the emotion words into the emotion dictionary to calculate the emotion values of modern Chinese literature texts.
- Prefab-Net: Segmentation of prefabricated building from satellite imagery with deep neural network–
A customized neural network structure is proposed, which combines multiple refined modules and improves the feature extraction, especially for the prefabricated building segmentation task, which proves the feasibility and superiority in practical application.
- –
- –
- –
- Automatic segmentation and recognition of red and white cells in stool microscopic images of human–
The experimental results indicate that the proposed marker-controlled watershed method can help increase the segmentation and recognition accuracy and is also less susceptible to the heteromorphic red and white cells.
- –
- The application of artificial immune algorithm to earthquake relief route planning–
A quick path planning method based on Artificial Immune Algorithm and helicopter operation performance in order to design an optimal path for an operational area of an extreme earthquake area is presented.
- –
- –
- Method of Key-Frame Selection for Video Images–
The experimental results show that the three-dimensional reconstruction of the selected image sequence as the source image reduces the number of feature points, shortens the time of feature matching, and improves the efficiency of 3D reconstruction by 3.6 times.
- Comparison of different anesthetic methods in the establishment of animal model of severe acute pancreatitis in rats–
It is better to choose the anesthetic method of inhaling ether when establishing SAP model in rats for its less influence on the some physiological indexes.
- AB051. Comparison of urine cytology and UroVysion FISH for upper tract urothelial carcinoma detection in Chinese patients–
Comparing with urine cytology, UroVysion is an efficient assay for upper tract urothelial carcinoma (UTUC) diagnosis and monitoring, with excellent clinical performance and utility in Chinese patients.
- –
- Robust foreground object segmentation from handheld camera videos with occlusion map–
A novel perspective to address partial movement and cast shadow in videos captured by a handheld camera by utilizing occlusion map introduced by object and camera motion and taking the advantage of interactive image segmentation methods.
- Dimensional coordinate measurements: application in characterizing cervical spine motion–
The layout of experimental system and a mathematical model to get the coordinates of dimensional coordinate of the marked points pasted to the human body’s cervical spine by an optical method based on binocular stereo vision are described.
- –
- –
- –
- –
- –
- –
- Mixture Analysis by Multichannel–
Experimental results demonstrate that the proposed multichannel structure makes the Hopfield-based mixture analysis feasible for real-world applica- tions with acceptable time cost and that the NEP stopping criterion can approach optimal unmixing results adaptively and accurately.
- –
- –
- –
- –
- Shape classification using hidden Markov model and structural feature–
A novel shape classification method based on Hidden Markov Models (HMMs) that employs HMMs to model the relationship among structural segments of the contour to demonstrate that the proposed method is efficient in classifying shapes, particularly unclosed shapes and similar shapes.
- –
- –
- Utilizing Structural Context for Region Classification–
Detailed experimental results demonstrate that the proposed approach to take structural context of image regions into account for region classification through a structural neural network has a great potential in region classification.
- –
- Control Design and Implementation in the Classroom Instruction Management Process–
A design for control over teaching activities is proposed, covering all possible problems in three stages of teaching by employing the theory of feed-forward, concurrent and feedback controls, along with a proposal concerning how to use it.
- –
- –
- Exploration on Setting up an Evaluation Norm System by Students for College Textbooks–
This research has set up an evaluation norm system by students for college textbooks that can reflect a basic picture of textbooks and is easy to be operated.
- –
- Ways to Use EXCEL to Process the Data of the Test Grade Forms and the Analysis Sheets–
This paper introduces ways to process the data of the test grade forms like the number in each grade by means of EXCEL, ways to calculate the ratio of pass, the average score, the standard deviation and ways to create a grade diagram.
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-15. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.