Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks105 from public data
- Improving 3D Object Detection with Channel-wise Transformer316
This paper leverages the high-quality region proposal network and a Channel-wise Transformer architecture to constitute the two-stage 3D object detection framework (CT3D) with minimal hand-crafted design and achieves superior performance and excellent scalability.
- Multi-Granularity Generator for Temporal Action Proposal216
Through temporally adjusting the segment proposals with fine-grained information based on frame actionness, MGG achieves the superior performance over state-of-the-art methods on the public THUMOS-14 and ActivityNet-1.3 datasets.
- 143
- Single infrared image enhancement using a deep convolutional neural network133
A new refined convolutional neural architecture is designed that produces visually very appealing results with higher contrast and sharper details compared to other network architectures for single infrared image enhancement task.
- Single Infrared Image Optical Noise Removal Using a Deep Convolutional Neural Network115
With a fully convolutional neural network, this method is able to eliminate the optical noise in single infrared image and significant image quality in experiments is achieved compared with the existing method.
- Thermal infrared colorization via conditional generative adversarial network111
This paper proposes a composite loss function that combines content, adversarial, perceptual and total variation losses that is used to recover global image information while the latter three losses are used to synthesize local realistic textures.
- Single Image Super-Resolution Using Compressive Sensing With a Redundant Dictionary95
This paper deals with the SR reconstruction problem based on the theory of compressive sensing, which uses a redundant dictionary instead of a conventional orthogonal basis and demonstrates better results on true images in terms of peak signal-to-noise ratio (PSNR) and root-mean-square error (RMSE) and give several important improvements, compared with other methods.
- Hyperspectral Image Super-Resolution via Dual-Domain Network Based on Hybrid Convolution73
A novel super-resolution (SR) algorithm for HSIs, named SRDNet, is proposed by using a dual-domain network with hybrid convolution and progressive upsampling to exploit both spatial-spectral and frequency information of the hyperspectral data.
- 69
- 54
- Improving Pixel-based MIM by Reducing Wasted Modeling Capability49
This paper is the first to systematically investigate multilevel feature fusion for isotropic architectures like the standard Vision Transformer (ViT) and is the first to systematically investigate multilevel feature fusion for isotropic architectures like the standard Vision Transformer (ViT).
- PixMIM: Rethinking Pixel Reconstruction in Masked Image Modeling47
A remarkably simple and effective method that entails filtering the high-frequency components from the reconstruction target to de-emphasize the network's focus on texture-rich details and adopting a conservative data transform strategy to alleviate the problem of missing foreground in MIM training is proposed.
- 45
- Adaptive Contrast Enhancement for Infrared Images Based on the Neighborhood Conditional Histogram42
Qualitative and quantitative evaluation results demonstrate that the proposed method outperforms the other block-based methods on local contrast enhancement, visual quality improvement, and noise suppression.
- Improved generative adversarial network and its application in image oil painting style transfer36
The experimental results show that the WGAN-GPalgorithm constructed in this study has the ability of stable gradient and alternating iterative convergence, and the total variation loss noise constraint can provide good edge and texture details for the migration process of image oil painting style.
- Mitigating and Evaluating Static Bias of Action Representations in the Background and the Foreground35
This paper empirically verify the existence of foreground static bias by creating test videos with conflicting signals from the static and moving portions of the video, and proposes a simple yet effective technique, StillMix, to learn robust action representations.
- FPN estimation based nonuniformity correction for infrared imaging system35
A novel scene-based nonuniformity correction method is proposed based on FPN estimation that leads to the less interference of the details in scenes and the temporarily still scenes, thus the proposed method performs well on the actual scene.
- Research on deep learning garbage classification system based on fusion of image classification and object detection classification30
Practice has proved that the recognition rate of garbage image classification has been increased to about 98%, and it has been transplanted to the raspberry pie microcomputer to achieve ideal results.
- JPEG image steganography payload location based on optimal estimation of cover co-frequency sub-image30
A JPEG image steganography payload location method based on optimal estimation of cover co- frequencies sub-image, which estimates the cover JPEG image based on the Markov model of co-frequency sub- image.
- 26
- Fast pyrolysis of holocellulose for the preparation of long-chain ether fuel precursors: Effect of holocellulose types26
The pyrolysis behaviors of nine biomass-derived holocelluloses were studied and the correlation of sugars/levoglucosan content with a mass ratio of cellulose to hemicellulose (CE/HCE) was put forward.
- Adaptive contrast enhancement based on histogram modification framework26
A novel mapping function is introduced to further improve the enhanced result of histogram equalization (HE) and get better performance on contrast enhancement and visual quality of the enhanced results.
- Optimized Contrast Enhancement for Infrared Images Based on Global and Local Histogram Specification26
Qualitative and quantitative evaluation results demonstrate that the proposed method outperforms the block-based methods for improving the visual quality of infrared images.
- Object Tracking Using Spatio-Temporal Networks for Future Prediction Location24
An object tracking algorithm that predicts the future locations of the target object and assists the tracker to handle object occlusion and sets new state-of-the-art performance on a few commonly used tracking benchmarks.
- 22
- Gradient domain weighted guided image filtering19
Experimental results demonstrate that the proposed algorithm significantly suppresses halo artifacts at the edges, making it highly effective for both image denoising and detail enhancement.
- MSRA atT TRECVID 2008: High-Level Feature Extraction and Automatic Search.19
This paper describes the MSRA experiments for TRECVID 2008, which representatively investigated the benefit of global and local low-level features by a variety of learning-based methods, including supervised and semi-supervised learning algorithms.
- 18
- 15
- Improved DV-Hop Localization Algorithm Based on Bat Algorithm in Wireless Sensor Networks15
An improved DV-Hop localization algorithm based on the bat algorithm (IBDV-Hop) based on optimization methods that enhance the accuracy of the average hop distance and fitness function and introduces a nonlinear dynamic inertial weight strategy to extend the global search scope and increase the local search accuracy.
- MARN: Multi-level Attentional Reconstruction Networks for Weakly Supervised Video Temporal Grounding14
- Boosting Video Description Generation by Explicitly Translating from Frame-Level Captions13
This paper proposes a novel sequence to sequence architecture to generate descriptions for videos, in a sense that the inputs are the captions of sequential frames and it outputs words sequentially.
- 12
- Automatic Lumbar Spine Tracking Based on Siamese Convolutional Network12
The study demonstrates that the lumbar tracker based on siamese convolutional network can be trained successfully without annotated lumbars sequences, and shown to track a previously unseen object without any adapting online.
- 11
- Fusion-Former: Fusion Features across Transformer and Convolution for Building Change Detection10
This approach fuses window-based self-attention with depth-wise convolution, which is named Fusion-Block and which combines convolutional neural networks (CNN) and a transformer to integrate information at different scales effectively to solve change detection challenges.
- Real-infraredSR: real-world infrared image super-resolution via thermal imager9
An infrared real-world dataset with different resolutions based on a refrigerated thermal detector and the infrared zoom lens is proposed, enabling the network to acquire more realistic details and an asymmetric residual block structure is proposed to effectively reduce the number of parameters and improve the performance of the network.
- 7
- 7
- 6
- Efficacy of cardioneuroablation for vasodepressor vasovagal syncope6
CNA is safe and effective in the treatment of vasodepressor VVS and provides a theoretical basis for individualization of treatment in patients with vasodepressor VVS.
- 6
- 6
- Data-driven methods for situation awareness and operational adjustment of sustainable energy integration into power systems6
An automatic adjustment model for power flow convergence using the D3QN (Double Dueling Q Network) reinforcement learning algorithm is introduced, addressing the challenge of power imbalance leading to flow non-convergence, enabling effective control of power flow converge and adaptive adjustment of operating modes.
- Object Tracking Based on Vector Convolutional Network and Discriminant Correlation Filters6
This paper proposes a novel visual tracking algorithm, namely VDCFNet, and combines DCF with a vector convolutional network (VCNN), which enables a model with few memories trained offline to learn the generic image features.
- 6
- 6
- 6
- 5
- 5
- Cross-domain colorization of unpaired infrared images through contrastive learning guided by color feature selection attention5
An innovative approach utilizing contrastive learning for unsupervised cross-domain mapping between unpaired infrared and visible color images, and introduces a color feature selection attention module guiding rational infrared image coloring.
- Automatic selection of landmarks in T1-weighted head MRI with regression forests for image registration initialization5
A learning-based method to automatically find a set of robust landmarks in three-dimensional MR image volumes of the head that are then used to compute a thin plate spline-based initialization transformation.
- 4
- 4
- Multi-Scale pattern-Aware task-Gating network for aerial small object detection3
A Multi-Scale Pattern-Aware Task-Gating Network (MPTNet) backbone that employs a small and large kernel convolutional complementary strategy to capture both large-scale and small-scale spatial patterns, yielding more comprehensive semantic features and a Multi-Head Cross-Space Encoder (MCE) that improves semantic fusion and spatial representation across hierarchical levels.
- Infrared and visible image fusion via pre-fusion compensation and visual-gradient saliency detection3
- Feature Selection of the Rich Model Based on the Correlation of Feature Components3
The results show that the proposed feature selection method can effectively reduce the dimensionalities of JPEG domain and spatial domain Rich Model steganalysis features without affecting the detection accuracies.
- 3D convolution network and Siamese-attention mechanism for expression recognition3
Experiments show that the proposed variable length 3D convolution network not only achieves comprehensive better performance than previous methods, but also have better generalization ability due to the attention mechanism.
- A Collaborative and Predictive Localization Algorithm for Wireless Sensor Networks3
The Gaussian mixture model (GMM) is introduced to predict the posterior trajectory for a mobile target by training its prior trajectory and results prove that the CPLA exhibits higher localization accuracy than other tested predictive localization algorithms either in the WSN with sparsely distributed sensor nodes or in theWSN with densely distributed sensor node nodes.
- Compressive Super-Resolution Imaging Based on Scrambled Block Hadamard Ensemble3
This paper proposes a novel model to achieve compressive super-resolution imaging based on an optical realization of scrambled block Hadamard ensemble sensing operation and demonstrates that the proposed model can achieve a four-time higher spatial resolution imaging with a small number of measurements in a single-shot acquisition process.
- Click-through-based Deep Visual-Semantic Embedding for Image Search3
A novel click-through-based deep visual-semantic embedding (C-DVSE) model for learning query and image similarity that is shown to be powerful for keyword-based image search with superior performance over several state-of-the-art methods.
- 3
- 2
- A Dual-Task Improved Transformer Framework for Decoding Lower Limb Sit-to-Stand Movement from sEMG and IMU Data2
iTransformer-DTL is presented, a dual-task learning framework with an improved Transformer designed to identify end-to-end locomotion modes and predict joint trajectories during sit-to-stand transitions that could support adaptable rehabilitation exoskeleton controllers, enhancing human–robot interactions.
- 2
- Coarse-to-fine Point Cloud Registration Based on Superpoint Overlap Prediction2
This paper proposes a coarse-to-fine point cloud registration method based on superpoint overlap prediction, which focuses on optimizing the construction of superpoint correspondences at the coarse scale and effectively improving registration accuracy.
- 2
- 2
- Object Recognition of Urban Buildings in High Spatial Resolution Remote Sensing Imagery2
The complementary advantages between image processing and deep learning are combined to research target-level extraction of high-resolution remote sensing urban buildings based on the building element information data and it is shown that the method proposed can be applied to the effective extraction of urban-level buildings.
- 2
- Siamese convolutional networks for tracking the spine motion2
Results indicate that the proposed tracking method can detect the lumbar vertebra steadily and robustly and can still achieve good tracking performance for images with low contrast and cluttered background.
- 2
- 2
- 2
- 1
- Dual-Branch Colorization Network for Unpaired Infrared Images Based on High-Level Semantic Features and Multiscale Residual Attention1
DC-Net is a dual-branch contrastive learning network that combines perceptual features and multiscale residual attention for the unsupervised cross-modal transformation of infrared to visible images, and demonstrates substantial improvements across all metrics.
- A lightweight model for action recognition based on depth separable convolution and LSTM1
Through the combination of a deep separable convolutional network and LSTM network, the proposed model has better expressive and generalization abilities, and is able to achieve better performance and results in the field of action recognition.
- Research on Pedestrian Detection Based on Multimodal Infor-mation Fusion1
1 × 1 convolution and expanded convolution square measure are introduced within the residual network, and the ROI Align methodology is employed to exchange the ROI Pooling method-ology to map the candidate box to the feature layer to optimize the Faster R-CNN.
- 1
- Find you wherever you are: geographic location and environment context-based pedestrian detection1
A context-based pedestrian detection approach is proposed by jointly exploring visual and contextual cues in a probabilistic model by constructing a contextual image database, in which each image is automatically attached with geographic location and environment information.
- 1
- 1
- Algorithm of optical remote sensing image registration based on strong edge region1
Aiming at the registration of optical remote sensing images, an algorithm based on strong edge region is proposed that combines with the regions' moment invariants and RANSAC method to obtain an accurate match of the strong edge regions.
- –
- –
- –
- –
- –
- Estimation of average heart rate and blood oxygen saturation via photoplethysmography analysis–
The optimization was intended to identify the minimum sampling rate and the minimum number of discrete data points at which the SpO2 accuracy could be secured, and revealed that with a sampling rate of 15 Hz and a collection period of 4.3 s, the SpO2 accuracy can be maintained at an acceptable level by 64-point FFT computation.
- –
- –
- –
- Research on Intelligent Recognition of Remote Sensing Images Based on U-Net Neural Network Algorithm–
- –
- Efficacy of cardioneurablation for vasodepressor vasovagal syncope–
This study provides a theoretical basis for individualization of treatment in patients with vasodepressor VVS, and concludes CNA is safe and effective in the treatment of vasodepsiderm VVS.
- MagnifierNet: Towards Semantic Regularization and Fusion for Person Re-identification.–
A novel network which accurately mines details for each semantic region and selectively fuse all semantic feature representations and introduces a novel loss function “Semantic Diversity Loss” (SD Loss) to facilitate feature diversity and improves regularization among all semantic regions.
- –
- Hierarchically grouped 2D local features applied to edge contour localisation–
A 2D object localisation system is proposed in this thesis that matches features of edge contour points, which is an alternative method that takes advantage of the shape information for objectLocalisation to avoid the need for linear contour detection and reduce the pathological disruption from the image background.
- Steganalysis of F5-like steganography based on selection of joint distribution features–
Experimental results show that the proposed feature selection and fusion method can achieve higher steganalytic accuracy and enhance the detection reliability.
- –
- –
- Application of Optimal-angle Transmission after 3D Reconstruction and Conventional DSA to Diagnosis of Intracranial Aneurysm Embolization–
DSA acquisition after optimal-angle transmission is better than conventional DSA in the display of intracranial aneurysm neck.
- Bolus Technology of GE INNOVA 3100 Digital Plate DSA Machine for Lower Extremity Angiography–
Bolus technology was valuable for the diagnosis of lower extremity arterial thrombosis and is better than traditional DSA in the aspect of display ofLower extremity vessel.
- –
- An improved arithmetic on FWM for WDM simulation software–
Analytical model to characterize fiber nonlinearities in the WDM Simulation Software have been applied and the simulation results of software are shown.
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.