Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks174 from public data
- Deep 3D human pose estimation: A review402
A thorough review of existing deep learning based works for 3D pose estimation is provided, the advantages and disadvantages of these methods are summarized, and the commonly-used benchmark datasets are explored for comparison and analysis.
- Crowd Counting and Density Estimation by Trellis Encoder-Decoder Networks366
This paper proposes a trellis encoder-decoder network (TEDnet) for crowd counting that achieves the best overall performance in terms of both density map quality and counting accuracy, and proposes a new combinatorial loss to enforce similarities in local coherence and spatial correlation between maps.
- 276
- Few-Shot Semantic Segmentation with Democratic Attention Networks241
This paper introduces the de-mocratized graph attention mechanism, which can activate more pixels on the object to establish a robust correspondence between support and query images, and proposes multi-scale guidance by designing a refinement fusion unit to fuse features from intermediate layers for the segmentation of the query image.
- Spatio-Temporal Laplacian Pyramid Coding for Action Recognition236
The proposed STLPC method achieves superb recognition rates on the KTH, the multiview IXMAS, the challenging UCF Sports, and the newly released HMDB51 datasets, and outperforms state of the art methods showing its great potential on action recognition.
- Learning to Learn with Variational Information Bottleneck for Domain Generalization206
A probabilistic meta-learning model for domain generalization is introduced, in which classifier parameters shared across domains are modeled as distributions, which enables better handling of prediction uncertainty on unseen domains.
- Memory Attention Networks for Skeleton-Based Action Recognition204
A newly temporal-then-spatial recalibration method named memory attention networks (MANs) is proposed and deploy MANs using the temporal attention recalibrration module (TARM) and spatiotemporal convolution module (STCM) to handle skeleton-based action recognition.
- 193
- 141
- 125
- 122
- 115
- Multi-scale deep networks and regression forests for direct bi-ventricular volume estimation115
A general, fully learning-based framework for direct bi-ventricular volume estimation, which removes user inputs and unreliable assumptions, and largely outperforms existing direct methods on a larger dataset of 100 subjects including both healthy and diseased cases with twice the number of subjects used in previous methods.
- Learning Discriminative Key Poses for Action Recognition99
A new classifier named weighted local naive Bayes nearest neighbor is proposed for the final action classification, which is demonstrated to be more accurate and robust than other classifiers, e.g., support vector machine (SVM) and naive Baye nearest neighbor.
- Conditional Variational Image Deraining91
A Conditional Variational Image Deraining (CVID) network is proposed for better deraining performance, leveraging the exclusive generative ability of Conditionalvariational Auto-Encoder on providing diverse predictions for the rainy image.
- Multi-Stream Convolutional Neural Network for SAR Automatic Target Recognition85
This paper proposes a novel deep convolutional learning architecture, called Multi-Stream CNN (MS-CNN), for ATR in SAR by leveraging SAR images from multiple views that can achieve high recognition rates and outperform other state-of-the-art ATR methods.
- 81
- Direct Estimation of Cardiac Bi-ventricular Volumes with Regression Forests77
With the proposed method, the most daily-used estimation of cardiac function, e.g., ejection fraction, can be conducted in a much more efficient, accurate and convenient way.
- Transductive Relation-Propagation Network for Few-shot Learning76
This work proposes a transductive relation-propagation graph neural network (TRPN) to explicitly model and propagate such relations across support-query pairs, the first work that explicitly takes the relations of support- query pairs into consideration in few-shot learning.
- Embedding Motion and Structure Features for Action Recognition72
This work proposes a novel method to model human actions by explicitly coding motion and structure features that are separately extracted from video sequences and integrates them in a unified framework, which provides an informative and compact representation of human actions.
- 67
- 66
- Knowledge-Aware Prompt Tuning for Generalizable Vision-Language Models63
This paper designs two complementary types of knowledge-aware prompts for the text encoder to leverage the distinctive characteristics of category-related external knowledge and designs an adaptation head for the visual encoder to aggregate salient attentive visual cues, which establishes discriminative and task-aware visual representations.
- 63
- Deep learning DCE-MRI parameter estimation: Application in pancreatic cancer58
This work investigated physics-informed deep neural networks for estimating physiological parameters from DCE-MRI signal-curves and selected the GRU framework for in vivo evaluation and compared to the spatio-temporal frameworks in 28 patients with pancreatic cancer.
- 55
- 54
- Action recognition by spatio-temporal oriented energies54
The unified STSP achieves comparable results with the state-of-the-art methods and takes advantage of multi-scale, multi-orientation analysis and feature pooling to produce a compact but informative and invariant representation of human actions.
- Direct and Simultaneous Four-Chamber Volume Estimation by Multi-Output Regression53
This paper proposes a new method for direct and simultaneous four-chamber volume estimation using multi-output regression that can disentangle complex relationship of image appearance and four- chamber volumes via statistical learning using a supervised descriptor learning SDL algorithm.
- Deep appearance and motion learning for egocentric activity recognition51
Experimental results show that the proposed deep appearance and motion learning (DAML) achieves high performance on both short- and long-term egocentric activity recognition tasks, which is comparable to or even better than the state-of-the-art counterparts.
- Action recognition via spatio-temporal local features: A comprehensive study50
A comprehensive study on local methods for human action recognition based on spatio-temporal local features, which implements these techniques and conducts comparison under unified experimental settings on three widely used benchmarks, i.e., the KTH, UCF-YouTube and HMDB51 datasets.
- 48
- Direct and simultaneous estimation of cardiac four chamber volumes by multioutput sparse regression46
A new, general framework for direct and simultaneous four chamber volume estimation by the multioutput sparse latent regression (MSLR), which enables jointly modeling nonlinear input‐output relationships and capturing four‐chamber interdependence is proposed.
- 43
- 43
- 42
- Local Feature Discriminant Projection42
This paper introduces the Differential Scatter Discriminant Criterion (DSDC) to the subspace learning of local feature descriptors which avoids the matrix singularity problem and proposes a generalized orthogonalization method to impose on projections, leading to a more compact and less redundant subspace.
- 41
- Uncertainty-aware report generation for chest X-rays by variational topic inference40
The results demonstrate that the proposed variational topic inference method can generate reports with novel sentence structure, rather than mere copies of reports used in training, while still achieving comparable performance to state-of-the-art methods in terms of standard language generation criteria.
- Seminar Learning for Click-Level Weakly Supervised Semantic Segmentation39
Seminar learning is proposed, a new learning paradigm for semantic segmentation with click-level supervision that achieves the new state-of-the-art performance of 72.51% (mIOU), surpassing previous methods by a large margin of up to 16.88% on the Pascal VOC 2012 dataset.
- 38
- 38
- Meta Learning to Bridge Vision and Language Models for Multimodal Few-Shot Learning37
The experimental results show that the multimodal meta-learning approach outperforms the baseline across multiple datasets and various training settings while being computationally more efficient.
- Variational Topic Inference for Chest X-Ray Report Generation36
The results demonstrate that the proposed variational topic inference method can generate novel reports rather than mere copies of reports used in training, while still achieving comparable performance to state-of-the-art methods in terms of standard language generation criteria.
- 36
- Combining appearance and structural features for human action recognition33
The results prove that the optimised 3D shape context offers complementary information to appearance features, showing its effectiveness for action representation; moreover, the proposed descriptor yields comparable results with the state-of-the-art methods.
- Few-Shot Ensemble Learning for Video Classification with SlowFast Memory Networks32
This paper introduces a family of few-shot learners based on SlowFast networks which are used to extract informative features at multiple rates, and incorporates a memory unit into each network to enable encoding and retrieving crucial information instantly.
- 31
- Real-time visual tracking based on improved perceptual hashing31
Experimental results show that the improved perceptual hashing-based tracking algorithms perform favorably against the state-of-the-art algorithms under various challenging environments in terms of time cost, accuracy and robustness.
- Long-Short-Term Features for Dynamic Scene Classification30
A unified framework to extract spatial and temporal features for dynamic scene representation is proposed and two variants of deep convolutional neural networks are deployed to encode spatial appearance and short- term dynamics into short-term deep features (STDF).
- Robust point pattern matching based on spectral context29
A local structural descriptor, namely the spectral context, is proposed to describe the attribute domain of point sets, which is fundamentally different from the previous methods, and the approximate distance order is defined and employed as the metric for geometric consistency of neighboring points in this work.
- 28
- 28
- Two-Stream Multi-Task Network for Fashion Recognition28
The proposed model achieves state-of-the-art results on large-scale fashion dataset comparing to the existing methods, which demonstrates its great effectiveness and superiority for fashion recognition.
- Attentional Alignment Networks.27
This work proposes attentional alignment networks (AAN), a novel end-to-end convolutional architecture for direct face alignment without relying on cascaded regression that consistently delivers high performance on all datasets, surpassing previous methods by large margins.
- Spatial and temporal scoring for egocentric video summarization27
This work utilizes egocentric salience cues, motion cues and a selection model to capture stable salience weight, discriminative weight and representative weight of a video shot respectively and combines these weights in a unified framework to predict the importance score of a shot.
- Cognitive Assessment Prediction in Alzheimer’s Disease by Multi-Layer Multi-Target Regression25
The multi-layer multi-target regression (MMR) is proposed which enables simultaneously modeling intrinsic inter-target correlations and nonlinear input-output relationships in a general compositional framework and has been evaluated by extensive experiments on the ADNI database with MRI data, and produced high accuracy surpassing previous regression models.
- 25
- Energy-Based Test Sample Adaptation for Domain Generalization24
A discriminative energy-based model is designed, which is trained on source domains to jointly model the conditional distribution for classification and data distribution for sample adaptation, and is optimized to simultaneously learn a classifier and an energy function.
- 23
- Variational Knowledge Distillation for Disease Classification in Chest X-Rays23
A conditional latent variable model is introduced, where the model acquires the ability to extract the visual features relevant to the disease during learning and can therefore perform more accurate classification for unseen patients at inference based solely on their X-ray scans.
- 23
- A local descriptor based on Laplacian pyramid coding for action recognition21
Experimental results on the benchmark KTH dataset and the newly released and challenging HMDB51 dataset demonstrate the effectiveness of the proposed 2D descriptor based on the Laplacian pyramid for human action recognition.
- Learning Hierarchical Embedding for Video Instance Segmentation20
This paper addresses video instance segmentation using a new generative model that learns effective representations of the target and background appearance and introduces normalizing flows to improve the robustness of the learned appearance embedding.
- SuperDisco: Super-Class Discovery Improves Visual Recognition for the Long-Tail18
This work proposes SuperDisco, an algorithm that discovers super-class representations for long-tailed recognition using a graph model, and learns to construct the super- class graph to guide the representation learning to deal with long-tails distributions.
- 18
- Order-preserving Consistency Regularization for Domain Adaptation and Generalization17
This work proposes the Order-preserving Consistency Regularization (OCR) for cross-domain tasks, which makes the model robust to task-irrelevant transformations and becomes less sensitive to the domain-specific attributes.
- Probabilistic Integration of Object Level Annotations in Chest X-ray Classification17
The proposed method yields consistent classification improvement across different back-bones on the common benchmark datasets Chest X-ray14 and MIMIC-CXR, showing how two-stage learning of labels from coarse to fine-grained, in particular with object level annotations, is an effective method for more optimal annotation usage.
- 17
- Attentive Temporal Pyramid Network for Dynamic Scene Classification17
The proposed attentive temporal pyramid network (ATP-Net) leverages the strength of attention mechanism to select the most relevant frame features and the ability of kernels to achieve optimal feature fusion for discriminative representations of dynamic scenes.
- Attentional prototype inference for few-shot segmentation16
This work proposes attentional prototype inference (API), a probabilistic latent variable framework for few-shot segmentation that obtains at least competitive and often better performance than state-of-the-art prototype-based methods.
- 16
- Heterogenous output regression network for direct face alignment15
A novel learning architecture, i.e., heterogenous output regression network (HORNet), for face alignment, which directly predicts facial landmarks from images, based on kernel approximations and establishes a new compact multi-layer architecture.
- 14
- MetaKernel: Learning Variational Random Features With Limited Labels14
This paper proposes meta-learning kernels with random Fourier features for few-shot learning by leveraging the shared knowledge provided by related tasks in a meta-learning setting and treats the random feature basis as the latent variable, which is estimated by variational inference.
- 12
- 12
- 11
- MetaModulation: Learning Variational Feature Hierarchies for Few-Shot Learning with Fewer Tasks11
The key idea is to use a neural network to increase the density of the meta-training tasks by modulating batch normalization parameters during meta- Training, and to account for the uncertainty caused by the limited training tasks, a variational MetaModulation is proposed.
- 11
- Model-Free Tracking With Deep Appearance and Motion Features Integration11
A real-time model-free object tracking approach is designed in this work relying on Convolutional Neural Networks (CNNs), which are deeply integrated by the proposed AMNet, which is an end-to-end offline trained two-stream network.
- 11
- Multi-task Shape Regression for Medical Image Segmentation11
A general segmentation framework of Multi-Task Shape Regression (MTSR) which formulates segmentation as multi-task learning to leverage its strength of jointly solving multiple tasks enhanced by capturing task correlations.
- 11
- 10
- Meta-Learning with Variational Semantic Memory for Word Sense Disambiguation10
This work proposes a model of semantic memory for WSD in a meta-learning setting that encapsulates prior experiences seen throughout the lifetime of the model, which aids better generalization in limited data settings and advances the state of the art in few-shot WSD.
- 10
- 10
- 10
- Handcrafted vs. learned representations for human action recognition10
Both feature learning and feature engineering have their advantages in visual representation which, however, remains less explored in the video domain for action recognition.
- Discriminative Embedding via Image-to-Class Distances10
A novel discriminative embedding method based on I2C for local feature dimensionality reduction that greatly reduces the computational burden and improves the performance of I 2C-based methods after reduction and provides an efficient closed-form solution by formulating the objective function as an eigenvector decomposition problem.
- Few-Shot Referring Video Single- and Multi-Object Segmentation Via Cross-Modal Affinity with Instance Sequence Matching9
A newly designed cross-modal affinity (CMA) module based on a Transformer architecture facilitates the establishment of multi-modal affinity over a limited number of samples, allowing the rapid acquisition of new semantic information while fostering the model’s adaptability to diverse scenarios.
- 9
- Multi-Constraint Transferable Generative Adversarial Networks for Cross-Modal Brain Image Synthesis9
A novel Brain Generative Adversarial Network (BrainGAN) that explores GANs with multi-constraint and transferable property for cross-modal brain image synthesis, and shows how BrainGAN can learn meaningful tissue representations with rich variability of brain images.
- 9
- 9
- 9
- Coarse-Fine View Attention Alignment-Based GAN for CT Reconstruction from Biplanar X-Rays8
This paper proposes a novel attention-informed coarse-to-fine cross-view fusion method to combine the features extracted from the orthogonal biplanar views that convey richer information than single-view X-rays and are more commonly used by surgeons.
- 8
- 8
- 8
- The Structure Transfer Machine Theory and Applications8
This paper proposes a new representation learning method, named Structure Transfer Machine (STM), which enables feature learning process to converge at the representation expectation in a probabilistic way and theoretically shows that such an expected value of the representation (mean) is achievable if the manifold structure can be transferred from the data space to the feature space.
- 8
- 8
- Direct volume estimation without segmentation8
Experimental results show that direct estimation methods provide more accurate estimation of cardiac ventricular volumes than segmentation based methods, which indicates thatdirect estimation methods not only provide a convenient and mature clinical tool for cardiac volume estimation but also enables diagnosis of cardiac diseases to be conducted in a more efficient and reliable way.
- LKA-ReID:Vehicle Re-Identification with Large Kernel Attention7
This paper proposes LKA-ReID with large kernel attention (LKA), which utilizes the advantages of self-attention and also benefits from the advantages of convolution, which can extract the global and local features of the vehicle more comprehensively.
- Classifying Normal and Abnormal Status Based on Video Recordings of Epileptic Patients7
A human action recognition scheme to detect distinct motion patterns and to distinguish the normal status from the abnormal status of epileptic patients is proposed and provides a fundamental analysis towards the human-robot interaction with socially assistive robots in caring the patients with epilepsy (or other patients with brain disorders) in order to protect them from injury.
- 6
- D3T: Dual-Domain Diffusion Transformer in Triplanar Latent Space for 3D Incomplete-View CT Reconstruction6
D3T is introduced, a new 3D conditional diffusion transformer that models 3D CT distributions in the low-dimensional 2D latent space for incomplete-view CT reconstruction and introduces a new Dual-Space Consistency Loss that integrates image-level supervision beyond standard supervision in the latent space to enhance consistency in the 3D image space.
- 6
- Reducing Fine-Tuning Memory Overhead by Approximate and Memory-Sharing Backpropagation6
The Approximate Backpropagation (Approx-BP) theory is proposed, which provides the theoretical feasibility of decoupling the forward and backward passes of backpropagation training, and a Memory-Sharing Backpropagation strategy, which enables the activation memory to be shared by two adjacent layers, thereby removing activation memory usage redundancy.
- Learning Cross-Modal Affinity for Referring Video Object Segmentation Targeting Limited Samples6
A simple yet effective model with a newly designed cross-modal affinity (CMA) module based on a Transformer architecture that adapts well to different scenarios with only a few samples, reaching state-of-the-art performance on the benchmark.
- Parameter-Free Channel Attention for Image Classification and Super-Resolution6
A Parameter-Free Channel Attention (PFCA) module is proposed to boost the performance of popular image classification and image super-resolution networks, but completely sweep out the parameter growth of channel attention.
- 6
- Correction to “Regression Segmentation for $M^{3}$ Spinal Images” [Aug 08 1640-1648]6
This paper aims to provide a history of the use of digital object Identifier in medical imaging and its applications in the clinical practice and aims to clarify its role in the diagnosis and treatment of central nervous system disorders.
- Introduction to Human Action Recognition6
This article provides an introduction to human action recognition and a comprehensive review on recent progress in both local and holistic representations of actions and describes the widely used benchmark human action datasets on which action recognition methods are evaluated and compared.
- 5
- 5
- 5
- 5
- Deep Collaborative Tracking Networks.5
The DCTN is the first generalized framework to model motion and appearance information with deep learning for object tracking and can achieve high tracking performance, which demonstrates the great effectiveness of exploring both motion and appearance information for visual object tracking.
- Human action representation using pyramid correlogram of oriented gradients on motion history images5
A novel descriptor, named pyramid correlogram of oriented gradients (PCOG), is presented for feature representation and shows that it gives more effective and powerful representation for human actions and can be used for the detection and recognition of a variety of actions.
- 4
- EMO: Episodic Memory Optimization for Few-Shot Meta-Learning4
EO is inspired by the human ability to recall past learning experiences from the brain's memory and is a simple plug-and-play optimizer that can be seamlessly embedded into existing optimization-based few-shot meta-learning approaches.
- Joint Super-Resolution and Inverse Tone-Mapping: A Feature Decomposition Aggregation Network and A New Benchmark4
A lightweight Feature Decomposition Aggregation Network (FDAN) is proposed that is efficient and outperforms state-of-the-art methods on joint SR-ITM and develops a Hierarchical FeatureDecomposition Group by cascading FDBs for powerful multi-level feature decomposition.
- 4
- 4
- GeneralizeFormer: Layer-Adaptive Model Generation Across Test-Time Distribution Shifts3
This work proposes to generate multiple layer parameters on the fly during inference by a lightweight meta-learned transformer, which it calls GeneralizeFormer: the layer-wise parameters are generated per target batch without fine-tuning or online adjustment.
- HYDEN: Hyperbolic Density Representations for Medical Images and Reports3
This work proposes a novel hyperbolic density embedding based image-text representation learning approach tailored for specific medical domain data, which integrates text-aware local features alongside global features from images, mapping image-text features to density features in hyperbolic space via using hyperbolic pseudo-Gaussian distributions.
- 3
- 3
- 3
- 3
- Feature Extraction and Representation for HumanAction Recognition3
This thesis aims to enhance human action recognition by feature extraction and representation using both holistic and local methods and proposes two distinctive algorithms for discriminative dimensionality reduction of local spatio-temporal descriptors.
- 2
- 2
- 2
- 2
- CageViT: Convolutional Activation Guided Efficient Vision Transformer2
Experimental results demonstrate that the proposed CageViT outperforms the most recent state-of-the-art backbones by a large margin in terms of efficiency, while maintaining a comparable level of accuracy.
- 2
- 2
- 2
- 2
- 1
- 1
- MetaTPT: Meta Test-time Prompt Tuning for Vision-Language Models1
Meta Test-Time Prompt Tuning is proposed, a meta-learning framework that learns a self-supervised auxiliary task to guide test-time prompt tuning and achieves state-of-the-art performance on domain generalization and cross-dataset benchmarks.
- Learning to Generalize Heterogeneous Representation for Cross-Modality Image Synthesis via Multiple Domain Interventions1
This work proposes a convolutional sparse coding (CSC) adaptation network to handle the lacking of generalizing medical image representation learning and reduces both inter-domain and intra-domain divergences by the domain-adaptation and domain-standardization modules.
- 1
- LKASeg:Remote-Sensing Image Semantic Segmentation with Large Kernel Attention and Full-Scale Skip Connections1
This paper proposes a decoder based on Large Kernel Attention (LKA), which extract global features while avoiding the computational overhead of self-attention and providing channel adaptability, and applies Full-Scale Skip Connections between the encoder and decoder.
- 1
- 1
- Cosine Activation in Compact Network (CACN): Application to Scene Classification1
The CACN is derived from kernel approximation and establishes a nonlinear hidden layer with the cosine activation function and is a well-directed solution to scene classification by seamlessly connecting with convolutional neural networks.
- 1
- –
- –
- –
- –
- Zero-shot Referring Image Segmentation Augmented by Multimodal Large Language Models–
A new framework is proposed, to leverage multimodal large language models (MLLM) into segmentation, and augmentations are considered from two aspects to align local image information to comprehensive linguistic information, under the hypothesis that the textual description guiding local target object in image is guiding local target object in image.
- –
- –
- –
- –
- Lightweight improved residual network for efficient inverse tone mapping–
A new Improved Residual Block (IRB) is proposed to extract and fuse multi-layer features for fine-grained HDR image reconstruction to enhance the power of popular residual block for efficient ITM.
- –
- Self-supervised Monocular Depth Estimation with Large Kernel Attention–
This paper proposes a decoder based on large kernel attention, which can model long-distance dependencies without compromising the two-dimension structure of features while maintaining feature channel adaptivity, and introduces a up-sampling module to accurately recover the fine details in the depth map.
- Reconstruct Spine CT from Biplanar X-Rays via Diffusion Learning–
This paper introduces an innovative method for 3D CT reconstruction utilizing biplanar X-rays that surpasses existing state-of-the-art benchmarks in both visual image quality and multiple evaluative metrics and incorporates a novel projection loss function.
- Self-Supervised Scene Flow Estimation with Point-Voxel Fusion and Surface Representation–
This paper proposes a point-voxel fusion method, where a voxel branch based on sparse grid attention and the shifted window strategy to capture long-range dependencies and a point branch to capture fine-grained features to compensate for the information loss in the voxel branch.
- –
- –
- –
- –
- –
- –
- –
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 2026-10-14. 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.