Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks226 from public data
- Single Image Dehazing via Multi-scale Convolutional Neural Networks1,759
A multi-scale deep neural network for single-image dehazing by learning the mapping between hazy images and their corresponding transmission maps by combining a coarse-scale net which predicts a holistic transmission map based on the entire image, and a fine-scale network which refines results locally.
- Deep People Counting in Extremely Dense Crowds388
This paper proposes an end-to-end deep convolutional neural networks (CNN) regression model for counting people of images in extremely dense crowds that achieves superior performance than the state-of-the-arts in term of the mean and variance of absolute difference.
- Single Image Dehazing via Multi-scale Convolutional Neural Networks with Holistic Edges382
A multi-scale deep neural network for single image dehazing by learning the mapping between hazy images and their transmission maps by combining a coarse-scale net which predicts a holistic transmission map based on the entire image, and a fine-scale network which refines dehazed results locally.
- 242
- 190
- 185
- 149
- 87
- 77
- 73
- 73
- 72
- 69
- Lightweight Pixel Difference Networks for Efficient Visual Representation Learning58
Two novel types of convolutions are proposed, dubbed Pixel Difference Convolution (PDC) and Binary PDC (Bi-PDC), which enjoy the following benefits: capturing higher-order local differential information, computationally efficient, and able to be integrated with existing DNNs.
- 56
- EPI-Patch Based Convolutional Neural Network for Depth Estimation on 4D Light Field54
An effective convolutional neural network framework to perform the depth estimation on 4-dimensional (4D) light field by based on the orientation-depth relationship of epipolar images (EPIs).
- Identity-Preserving Face Anonymization via Adaptively Facial Attributes Obfuscation53
A face anonymization framework that could obfuscate visual appearance while preserving the identity discriminability and can achieve a balance between recognition utility and appearance anonymizing by modifying different numbers of facial attributes according to pratical demands.
- 49
- 47
- 47
- 47
- 46
- 46
- 45
- 45
- 43
- Improving Dynamic HDR Imaging with Fusion Transformer42
This paper proposes a transformer model for HDR imaging that outperforms the state-of-the-art by large margins on several popular public datasets, and demonstrates that the key component is the HDR transformer module.
- Energy-Optimal Planning of Robot Trajectory Based on Dynamics42
It is demonstrated that the energy consumption of the trajectory is significantly reduced compared with that of the quintic polynomial trajectory and an obvious 82.3% decrease of energy consumption is obtained in the jerk of the optimal energy trajectory, which makes the manipulator to smoothly run.
- 42
- 40
- 36
- 36
- 34
- Human Action Recognition Via Multi-modality Information34
The proposed pyramid appearance and global structure action descriptors on both RGB and depth motion history images and a model-free method for human action recognition outperforms the state-of-the-art methods on both MSR Action3D and DHA datasets.
- Semantic Segmentation of High-Resolution Remote Sensing Images with Improved U-Net Based on Transfer Learning33
An improved U-Net model based on transfer learning to solve the semantic segmentation problem of high-resolution remote sensing images is proposed, based on the symmetric encoder–decoder structure of U- net.
- 33
- 33
- 32
- A Novel Action Saliency and Context-Aware Network for Weakly-Supervised Temporal Action Localization31
- 29
- 29
- 29
- 29
- Audio Visual Attribute Discovery for Fine-Grained Object Recognition29
This paper introduces a novel feature named audio visual attributes via discovering the correlations between the visual and audio representations and proposes a unified framework for training with video-level category label which can be implemented end-to-end in the step of inference.
- 28
- Toward Generalized Few-Shot Open-Set Object Detection25
A new G-FOOD algorithm is proposed, named Few-shOt Open-set Detector (FOOD), which contains a novel class weight sparsification classifier (CWSC) and a novel unknown decoupling learner (UDL) that can be identified with a confidence probability without any threshold, prototype, or generation.
- Exploring the Cross-Domain Action Recognition Problem by Deep Feature Learning and Cross-Domain Learning25
Deep feature learning method has much better generalization ability than the hand-crafted feature, such as improved dense trajectory if there are enough labeled samples in the training dataset to be used to fine-tune the network, and both unsupervised cross-domain learning method and supervised cross- domain learning method can improve the performance.
- Partial multi-label learning with noisy side information23
This work introduces a novel partial multi-label learning with noisy side information approach, which simultaneously removes noisy outliers from the training instances and trains robust partialmulti-label classifier for unlabeled instances prediction.
- 23
- 22
- Sim2Word: Explaining Similarity with Representative Attribute Words via Counterfactual Explanations21
A new interpretation method is proposed that explains the image similarity models by salience maps and attribute words and can be applied to evidential learning cases, such as finding the most characteristic attributes in a set of face images, and is verified on the VGGFace2 dataset.
- 20
- A comparison of the biological, geological events and environmental backgrounds between the Neoproterozoic-Cambrian and Permian-Triassic transitions20
A detailed comparison of events during the two critical transitions indicates that despite their general comparability, significant differences do exist in magnitude, duration, and frequency.
- Generalized Semantic Contrastive Learning via Embedding Side Information for Few-Shot Object Detection19
This work proposes a generalized feature learning method for FSOD by leveraging side information, and introduces a side-information guided region-aware masking module that improves sample diversity by removing biased information through counterfactual explanations.
- A Trajectory Regression Clustering Technique Combining a Novel Fuzzy C-Means Clustering Algorithm with the Least Squares Method19
This work defines the new concept of trajectory clustering and construct a novel partitioning (angle-based partitioning) method of line segments and employs least squares regression model to achieve regression clustering of the trajectory.
- DomainPlus: Cross Transform Domain Learning towards High Dynamic Range Imaging17
This paper proposes a cross-transform domain neural network for efficient HDR imaging, and proposes fronted Discrete Wavelet Transform and Discrete Cosine Transform-based learnable bandpass filters to formulate aCross- transform domain learning block, dubbed DomainPlus Block (DPB) for effective ghosting removal.
- 17
- Action recognition using 3D DAISY descriptor17
A recent image local descriptor, DAISY, is extended to three dimensions to deal with the information in the additional temporal domain in videos to create a novel spatial-temporal descriptor for action recognition.
- 16
- Deep Spatial Pyramid Features Collaborative Reconstruction for Partial Person ReID16
This paper proposes a novel spatial pyramid feature construction approach (DCR) for partial person ReID, which can effectively and efficiently tackle the occlusion in arbitrary sizes and outperforms all the competitors with a large margin.
- As‐global‐as‐possible stereo matching with adaptive smoothness prior16
The proposed AGAP outperforms other four most challenging stereo matching algorithms, and also performs better on Microsoft i2i stereo videos, and under various strategies of parallelisation shows a near real-time execution time.
- 15
- A Wavelet Neural Network for SAR Image Segmentation15
The WNN combines the multiscale analysis ability of the wavelet transform and the classification capability of the artificial neural network by setting the wave let function as the transfer function of the neural network.
- HSIC-based Moving Weight Averaging for Few-Shot Open-Set Object Detection14
To suppress the over-fitting problem and improve the model's generalization ability for unknown rejection, the HSIC-based (Hilbert-Schmidt Independence Criterion) moving weight averaging to update the weights of classification and regression heads, which considers the degree of independence between the current weights and previous weights stored in the long-term memory banks is proposed.
- Partial label learning via low-rank representation and label propagation14
This paper proposes a novel two-stage method called partial label learning via low-rank representation and label propagation, where instance similarity and label confidence are taken into consideration to improve the disambiguation ability of the model.
- 14
- 14
- Seeing is Not Believing: An Identity Hider for Human Vision Privacy Protection13
An effective identity hider for human vision protection, which can significantly change appearance to visually hide identity while allowing identification for face recognizer while allowing identification for face recognizers is proposed.
- Semantic image segmentation with fused CNN features12
This paper fuse the hierarchical convolutional neural network features and the region-based features as the feature representation and uses a fully connected conditional random field (CRF) as a post-processing method to improve the labeling consistency.
- 12
- 11
- 11
- Visual Tracking Based on an Improved Online Multiple Instance Learning Algorithm11
An improved online multiple instance learning (IMIL) for a visual tracking algorithm that is robust to occlusion and appearance changes and compared with other state-of-the-art algorithms is proposed.
- 10
- Image Retrieval and Ranking via Consistently Reconstructing Multi-attribute Queries10
This work discovers the correlation among attributes via expanding the query representation, and imposing the group sparsity on representations to reduce the disturbance of noisy data.
- LSL-SS-Net: level set loss-guided semantic segmentation networks for landslide extraction9
This study presents a level set loss-guided semantic segmentation network, which can be integrated with different semantic segmentation networks, to address issues of level set loss in various semantic segmentation networks and in different study areas for landslide extraction.
- 9
- 9
- 9
- 9
- FaceInsight: A Multimodal Large Language Model for Face Perception8
FaceInsight is proposed, a versatile face perception MLLM that provides fine-grained information and introduces visual textual alignment of facial knowledge to model both uncertain dependencies and deterministic relationships among facial information, mitigating the limitations of language-driven reasoning.
- Adaptive coding unit size convolutional neural network for fast 3D-HEVC depth map intracoding8
This work proposes an adaptive CU size CNNs for fast 3D-HEVC depth map intracoding, and employs spatial pyramid pooling to fully extract the features of the three types of CUs, and applies the nonlocal self-attention mechanism to make it suitable for depth maps.
- Real-Time 3D Hand Gesture Detection from Depth Images8
An real-time algorithm to detect 3D hand gestures from depth images by detecting moving regions by frame difference, then regions are refined by removing small regions and boundary regions, and foremost region is selected and its trajectories are classified using an automatic state machine.
- 8
- 7
- 7
- MMA: a multi-view and multi-modality benchmark dataset for human action recognition7
This paper introduces a multi-view, multi-modality benchmark dataset for human action recognition (abbreviated to MMA), and provides the baseline for the evaluation of existing state-of-the-art algorithms on MMA.
- 7
- 6
- 6
- 6
- A Novel 4D Track-before-Detect Approach for Weak Targets Detection in Clutter Regions6
A 4D TBD approach is developed here for closely weak extended target tracking and overcoming heterogeneous clutter background and various clutter regions, and it is proved that the method works well in tracking closely maneuvering extended targets even if a clutter region exists.
- 6
- A Novel RNN-GBRBM Based Feature Decoder for Anomaly Detection Technology in Industrial Control Network6
A novel method of anomaly detection in industrial control network is proposed based on RNN-GBRBM feature decoder based on a semi-supervised incremental updating algorithm in order to improve the performance of the model continuously.
- 6
- 6
- 6
- 5
- 5
- 5
- Magic pencil5
A novel image synthetic approach named Magic Pencil, which aims to synthesize the real images based on the sketch by designing a novel multi-task neural network based on generative adversarial nets composed of three components i.e. the generator, the discriminator, and the classifiers.
- Transparent Discovery of Hidden Service5
This letter proposes a novel approach that can transparently discover the client of the hidden service, based on extensive analysis on the hidden service protocol, and a combination of cells which can be used to generate a special traf-fic feature with the cell-padding mechanism of Tor.
- 5
- Approach of context-aware computing with uncertainty for ubiquitous active service5
This work modified the fusion method of evidence information after considering context's reliability, time-efficiency and relativity, which has improved the classic combination rule of D-S evidence theory when used in the pervasive computing paradigm and overcome its shortcoming.
- Two-Person Non Zero-Sum Bimatrix Game with Fuzzy Payoffs and Its Equilibrium Strategy5
This paper considers fuzzy bimatrix games with fuzzy payoffs based on fuzzy max order, and defines three kinds of concepts of minim ax equilibrium strategies for such games.
- 4
- A multi-level feature integration network for image inpainting4
This work proposes a multi-level feature integration (MFI) network for image inpainting, and uses PatchGAN to support the refinement generator network to produce more discriminative detail and demonstrates the effectiveness of this network for producing visually pleasing results with semantically ordered textures.
- 4
- 4
- Optimization and Simulation of Garden Image Visual Effect Based on Particle Swarm and Wavelet Threshold4
The wavelet basis function is adopted as the activation function of the deep belief network to improve the recognition accuracy of thedeep belief network, especially the Recognition accuracy of small changes in the garden image information.
- Fusion of multiple features to produce a segmentation algorithm for remote sensing images4
Experimental results show that the proposed method can decrease over-/under-segmentation and thus produce satisfactory segmentation results.
- 4
- 4
- Context constrained-generalized posterior probability for verifying phone transcriptions4
A new statistical confidence measure, Context ConstrainedGeneralized Posterior probability (CC-GPP), is proposed for verifying phone transcriptions in speech databases by considering string hypotheses that bear a focused phone with partially matched left and right contexts.
- Modern Education Technology and Foreign Language Teaching Mode4
It has been proved that modern education technology has incomparable advantage over the traditional education mode and therefore the initiative, the elicitation and the exhibition of modern educationtechnology bears important and imminent significance in the reform of foreign language teaching mode.
- Anti-Fake Vaccine: Safeguarding Privacy Against Face Swapping via Visual-Semantic Dual Degradation3
A novel privacy-preserving framework, named Anti-Fake Vaccine, to protect the facial images against the malicious face swapping, and develops an additive perturbation strategy to discover the shared adversarial perturbations across diverse face swapping models.
- 3
- 3
- 3
- 3
- Grid self-occlusion: a grid self-occlusion data augmentation for better classification3
A novel data augmentation method named grid self-occlusion, which exploits binary mask with inconsistent intervals to achieve the erasing balance of information and demonstrates the effectiveness of the proposed method.
- 3
- A Nonlocal Method with Modified Initial Cost and Multiple Weight for Stereo Matching3
Experimental results show that the proposed method outperforms the classical nonlocal methods for stereo matching and is a more robust and reasonable tree structure.
- 2
- 2
- 2
- A Low-Complexity Accurate Ranging Algorithm for a Switch Machine Working Component Based on the Mask RCNN2
A low-complexity accurate ranging algorithm of the Mask RCNN is proposed to measure the on–off working parts of turnout maintenance and can accurately measure the distance of different working parts.
- Grier: graph repairing based on iterative embedding and rules2
Grier is proposed, a novel repairing method that iteratively learns the graph embeddings with guidance from logical rules, which has significant power in knowledge acquisition and inference.
- Image Generation with Global Photographic Aesthetic Based on Disentangled Generative Adversarial Network2
This study uses an existing generative model, StyleGAN, to build the height of image content and puts forward a new method based on the GAN disentangled representation of a global aesthetic image generation algorithm by mining GANs’ latent space, potential global aesthetic feeling, and aesthetic editing of the original image to realize the aesthetic feeling and content of high-quality global aestheticimage generation.
- 2
- Analyzing Knowledge Innovation mechanism Using Patent Data Mining and Network construction2
This paper constructs individual cooperation network and enterprise mobility network by mining patent data of American biotechnology industry by proposing proposed approach to big data mining and measure of network.
- Adaptive LOD representation of terrain model based on quad-tree2
The experimental results show that when the drawing scene is large, the algorithm can greatly reduce the amount of calculation, improve the rendering efficiency of the terrain model, and ensure a good experience of the entire scene.
- 2
- Human Behavior Recognition Based on Axonometric Projections and PHOG Feature2
This paper proposes an effective method to recognize human behaviors from sequences of depth maps, which provide additional body shape and motion information for behavior recognition.
- 2
- 2
- Weber’s Law Based Center-Surround Hypothesis for Bottom-Up Saliency Detection2
Comparison experiments demonstrate this detector not only can generate high quality saliency maps with the same resolution as the input image, but also has stronger response in activation regions and better inhibition performance in other regions.
- Research and implementation of Course Teaching-Learning Process Management System2
This paper designs and implements a generic, complete Course Teaching-Learning Process Management System, it provides a flexible tracking and controlling of teaching-learning process and an effective management of teaching resources.
- 1
- Beyond Progress Measures: Theoretical Insights into the Mechanism of Grokking1
This work investigates the grokking mechanism underlying the Transformer in the task of prime number operations and provides a unified perspective for understanding various previously proposed progress measures and introduces a novel, concise, and effective progress measure that could trace the changes in test loss more accurately.
- 1
- PhaseWin Search Framework Enable Efficient Object-Level Interpretation1
A novel phase-window search algorithm that enables faithful region attribution with near-linear complexity and establishes a new state of the art in scalable, high-faithfulness attribution for object-level multimodal models.
- Spatiotemporal correlation enhanced real-time 4D-CBCT imaging using convolutional LSTM networks1
The spatiotemporal correlation-based respiration motion modeling supplied a potential solution for accurate real-time 4D-CBCT reconstruction and outperformed other networks in terms of visual inspection and quantitative metrics.
- 1
- 1
- 1
- 1
- RETRACTED: Target recognition and localization based on lightweight single-shot multibox detector network for robotics1
Based on the digital normalization and residual network structure, the depth-wise separable convolution is proposed to replace the traditional convolution in the field of intelligent inspection robots and intelligent security robots.
- 1
- 1
- LEAF1
A joint learning objective function which is composed of the high-order representation mining term and the classifier training term that can both preserve category-specific information and discover the common structure between categories.
- An Improved Genetic Algorithm for Task Scheduling in Distributed Computing System1
An improved genetic algorithm called genetic algorithm with geographical isolation is proposed, which restrains the excessive growth phenomenon of individuals by dividing the population, effectively solves the problem of premature convergence in genetic algorithm.
- 1
- Wind speed forecasting model based on wavelet decomposition and support vector machine1
WD-SVM model, a wind speed prediction model that uses data mining techniques of wavelet analysis and support vector machine, is developed and applied to a wind farm for a forecast of its 10-minute average wind speed four hours in advance.
- An Efficient Region Matching Method for Content-based Image Retrieval1
A novel region matching method for contentbased image retrieval where the region dependences in each image are first computed in order to select the most interesting region from each image.
- 1
- Feature Extraction in Speechreading1
Experimental results showed that this algorithm is superior to any other appearance-based feature extraction algorithm in speechreading, and specifically LDAO is better than DCT+LDA about 3%.
- 1
- 1
- A New Fractal Interpolation Approach for Geographic Curve1
A new fractal interpolation is presented, based on the fractal theory, Douglas Peuker algorithm and random midpoint displacement method, which has some advantages over the traditional interpolation methods.
- 1
- AI-Enabled Medical Education: Current Insights and Future Prospects–
Constructing a new human-machine collaborative educational ecology, improving ethical norms and evaluation systems, and promoting the creative reconstruction of teachers' roles are key pathways to ensure that AI technology truly empowers medical education and cultivates outstanding medical talents.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Exploration of Teaching Reform of Computer Network Practice Course–
The research focuses on the following reforms of computer network practice course teaching reform.
- –
- –
- –
- –
- –
- –
- An Information Extracting Scheme for Netdisk–
A method to extract downloading links of netdisk and another method based on the Drop Fall Algorithm to undertake the recognition of merged characters is proposed, which is suitable for the netdisk environment.
- –
- –
- Application Research of Multi-Software Collaboration Based on Complex Surface Design–
The research on using the collaboration of some different software to solve the problems and realize the seamless cohesion with the help of the respective special features of variety of CAD software and the compatibility of data interface has proved that the approach is a qualitative improvement.
- –
- –
- Bayesian Network Based Expert System for Seismic Performance Assessment–
This paper shows that the Bayesian network based expert system can perform bidirectional inference based on incomplete information when the prior probability distributions are given.
- Resource Sharing and Reference Service–
For the development of internet technology and abundant resource of library, to provide faster and better service to readers has become possible.
- Two-dimensional Teaching Animation Design Based on Cognitive Load Theory–
The factors influencing two-dimensional teaching animation from the perspective of cognitive load theory as well as the design principles for the effective control of its cognitive load are analyzed.
- HP Smart Print–
HP SmartPrint, a novel web browser plug-in which automatically suggests print-worthy content within a web page and provides an intuitive UI for users to make corrections to the initial suggestion, if needed, is presented.
- –
- –
- Evidence-based Medical and Library Service–
The libraries should know the information source of the evidence~based medicine searching, screen and distinguish for the users, so as to play the library society values.
- –
- –
- Practice and Study on Maintenance of Databases of College Libraries in Jiangsu Province–
The existing problems of the catalogue data are probed, the objective causes of the problems are analyzed as well as the specific solutions are provided.
- On Four-way Integration Teaching Model of Intensive English–
Four-way integration teaching model of intensive English, which integrates the traditional classroom teaching and e-teaching, optimizes students’ autonomous learning strategy and teaching evaluation system, develops students' integrated English ability.
- –
- –
- A new exemplar-based image completing method–
A new algorithm for exemplar-based image completion is proposed using color ratio gradient (CRG), which is robust to object occlusion and clustering which induce mismatch in general image matching algorithm.
- Method and analysis for assimilated processing of sensor measurement error distribution model–
A truncated triangular probability-possibility transformation to be applied to any unimodal and symmetric probability distribution was proposed and a fuzzy model of sensors measurement date was built.
- –
- –
- –
- Accelerate the Development and Utilization of Periodical Database–
The necessity of accelerating periodical database development and utilization is drawn out, and solutions and countermeasures from the collection, management and utilization of theperiodical database are put forward.
- –
- Speech endpoint detection based on lip moving–
Experiment shows that speech endpoint detection based on lip moving is effective and classify and recognize lip based on Probabilitic neural networks.
- –
- –
- The Method of Collocation Based on the Active Contour Model–
A method of high precision collocation based on the B-spline Snake model is advanced that can achieve the precise collocation between the image and the vectorgraph.
- –
- Research in Design of Digital Wave Filter Based on Matlab–
The FIR digital filter is designed based on the disposed function of digital signal on the DSP chip TMS320C5402 with the benefit of Matlab programming language and is being used in the disposed digital signal sphere widely.
- –
- Application of digital signal processor TMS320F2812 in design of intelligent release–
The function and working principle of intelligent release based on key processor TMS320F2812DSP is introduced and the key design technology of hardware and software are given,employing FFT arithmetic to deal with and analyse the harmonic.
- A Multiscale Self-growing Probabilistic Decision-Based Neural Network for Segmentation of SAR Imagery–
A new segmentation algorithm for synthetic aperture radar (SAR) image is proposed using multiscale self-growing probabilistic decision-based neural network (MSPDNN), able to find the natural number of category in SAR image based on the Bayesian information criterion (BIC).
- –
- –
- Implementation mechanism of virtual functions in C++ programming language–
The virtual function in C+G-+ is dynamically bound through a virtual pointer(vptr) and virtual table during the running time, and so different class virtual function are called by the object′s type.
- Resarch on the Construction Method of User Profiles for Decision Support Oriented Personal Assistant Agents–
The concept and representation of the user profiles of the personal assistant agents and the approach for acquiring user preferences are proposed, as the core of user profiles construction.
- Views on Periodical Registration–
It is shown that only when one tries his be st, can he brings into full play the role of periodicals, and it is put forward corresponding solution methods.
- –
- –
- Web-based English Reading Course--On Inquiry Learning–
The mode of inquiry learning is adopted in the web-based English reading course, which requires students to seek references to the text contents through the web, the on-line library, the campus data base, etc, to develop information literacy and inquiry learning ability while improving their English reading.
- A Design of Transformer Station Realtime Monitoring System Using Visual Basic–
This paper presents a design of a practical transformer station realtime monitoring system that involves serval key techniques such as serial communication, database, dynamic-report and so on.
- A NEW DSS BASED ON INTRANET AND DATA WAREHOUSE–
A new DSS based on Intranet and Data Mining was supposed and comparied with general DSS and it′s advantages were analysised.
- –
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.