Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks317 from public data
- Visual saliency based on multiscale deep features1,361
This paper discovers that a high-quality visual saliency model can be learned from multiscale features extracted using deep convolutional neural networks (CNNs), which have had many successes in visual recognition tasks.
- 920
- Neural Style Transfer: A Review895
A taxonomy of current algorithms in the field of NST is proposed and several evaluation methods are presented and compared to compare different NST algorithms both qualitatively and quantitatively.
- nnFormer: Volumetric Medical Image Segmentation via a 3D Transformer812
nFormer is introduced, a 3D transformer for volumetric medical image segmentation that not only exploits the combination of interleaved convolution and self-attention operations, but also introduces local and global volume-based self-Attention mechanism to learn volume representations.
- Deep Contrast Learning for Salient Object Detection774
This paper proposes an end-to-end deep contrast network that significantly improves the state of the art in salient object detection and extracts segment-wise features very efficiently, and better models saliency discontinuities along object boundaries.
- Mesh editing with poisson-based gradient field manipulation682
A novel approach to mesh editing with the Poisson equation as the theoretical foundation that modifies the original mesh geometry implicitly through gradient field manipulation, and can outperform previous related mesh editing techniques.
- Efficient View-Dependent Image-Based Rendering with Projective Texture-Mapping621
This paper presents how the image-based rendering technique of view-dependent texture-mapping (VDTM), a technique for generating novel views of a scene with approximately known geometry making maximal use of a sparse set of original views, can be efficiently implemented using projective texture mapping, a feature commonly available in polygon graphics hardware.
- Visual Saliency Detection Based on Multiscale Deep CNN Features394
This paper discovers that a high-quality visual saliency model can be learned from multiscale features extracted using deep convolutional neural networks (CNNs), which have had many successes in visual recognition tasks.
- Visual Saliency Detection Based on Multiscale367
This paper discovers that a high-quality visual saliency model can be learned from multiscale features extracted using deep convolutional neural networks (CNNs), which have had many successes in visual recognition tasks.
- A transformer-based representation-learning model with unified processing of multimodal input for clinical diagnostics349
The unified model outperformed an image-only model and non-unified multimodal diagnosis models in the identification of pulmonary disease and in the prediction of adverse clinical outcomes in patients with COVID-19 (by 29% and 7%, respectively).
- Adaptive Context Selection for Polyp Segmentation344
This work proposes an adaptive context selection based encoder-decoder framework which is composed of Local Context Attention (LCA) module, Global Context Module (GCM) and Adaptive Selection Module (ASM), which shows outstanding performance compared with other state-of-the-art methods.
- 337
- High-Resolution Shape Completion Using Deep Neural Networks for Global Structure and Local Geometry Inference331
This work proposes a data-driven method for recovering missing parts of 3D shapes based on a new deep learning architecture consisting of a global structure inference network and a local geometry refinement network that outperforms existing state-of-the-art work on shape completion.
- Weakly Supervised Complementary Parts Models for Fine-Grained Image Classification From the Bottom Up317
This paper builds complementary parts models in a weakly supervised manner to retrieve information suppressed by dominant object parts detected by convolutional neural networks and builds a bi-directional long short-term memory (LSTM) network to fuze and encode the partial information of these complementary parts into a comprehensive feature for image classification.
- Automatic Photo Adjustment Using Deep Neural Networks302
This article forms automatic photo adjustment in a manner suitable for deep neural networks, and introduces an image descriptor accounting for the local semantics of an image that can model local adjustments that depend on image semantics.
- Instance-Level Salient Object Segmentation297
This paper presents a salient instance segmentation method that produces a saliency mask with distinct object instance labels for an input image, and proposes a multiscale saliency refinement network, which generates high-quality salient region masks and salient object contours.
- Cross-modality deep feature learning for brain tumor segmentation292
The proposed cross-modality deep feature learning framework can effectively improve the brain tumor segmentation performance when compared with the baseline methods and state-of-the-art methods.
- Dynamic Graph Attention for Referring Expression Comprehension263
This paper proposes a dynamic graph attention network to perform multi-step reasoning by modeling both the relationships among the objects in the image and the linguistic structure of the expression, and constructs a graph for the image with the nodes and edges corresponding to the objects and their relationships respectively.
- 256
- Borrowing Treasures from the Wealthy: Deep Transfer Learning through Selective Joint Fine-Tuning256
This paper introduces a deep transfer learning scheme, called selective joint fine-tuning, for improving the performance of deep learning tasks with insufficient training data, and can improve the classification accuracy by 2% - 10% using a single model.
- AnL1image transform for edge-preserving smoothing and scene-level intrinsic decomposition242
An image transform based on the L1 norm for piecewise image flattening that can effectively preserve and sharpen salient edges and contours while eliminating insignificant details, producing a nearly piecewise constant image with sparse structures is introduced.
- A deep learning approach to characterize 2019 coronavirus disease (COVID-19) pneumonia in chest CT images221
The algorithm showed excellent performance in detecting COVID-19 pneumonia on chest CT images compared with resident radiologists, and the higher sensitivity of deep learning model in detectingCOVID- 19 pneumonia were found compared with radiological residents on a per-lobe and per-patient basis.
- Recovering photometric properties of architectural scenes from photographs220
The goal of generating photorealistic renderings of real architectural scenes under novel lighting conditions has been achieved and the basic approach to solving a series of optimization problems to find the parameters of appropriate lighting and reflectance models that best explain the measured values in the various photographs of the scene.
- Motion Guided Attention for Video Salient Object Detection216
A multi-task motion guided video salient object detection network, which learns to accomplish two sub-tasks using twoSub-networks, one sub-network for salient object Detection in still images and the other for motion saliency detection in optical flow images, which significantly outperforms existing state-of-the-art algorithms on a wide range of benchmarks.
- 214
- 212
- Multi-evidence Filtering and Fusion for Multi-label Classification, Object Detection and Semantic Segmentation Based on Weakly Supervised Learning210
A novel weakly supervised curriculum learning pipeline for multi-label object recognition, detection and semantic segmentation, and a novel algorithm for filtering, fusing and classifying object instances collected from multiple solution mechanisms is proposed.
- A clinically applicable deep-learning model for detecting intracranial aneurysm in computed tomography angiography images207
A deep-learning-based framework is presented achieving improved performance compared to that of radiologists and expert neurosurgeons for intracranial aneurysm diagnosis and it is found that 99.0% predicted-negative cases can be trusted with high confidence, leading to a potential reduction in human workload.
- EdgeStereo: An Effective Multi-task Learning Network for Stereo Matching and Edge Detection204
An effective multi-task learning network, EdgeStereo, composed of a disparity estimation branch and an edge detection branch, which enables end-to-end predictions of both disparity map and edge map and achieves comparable generalization performance for disparity estimation because of the incorporation of edge cues.
- A Survey on Graph Neural Networks and Graph Transformers in Computer Vision: A Task-Oriented Perspective194
This paper presents a comprehensive review of GNNs and graph Transformers in computer vision from a task-oriented perspective and divides their applications into five categories according to the modality of input data, i.e., 2D natural images, videos, 3D data, vision + language, and medical images.
- Few-shot medical image segmentation using a global correlation network with discriminative embedding194
This study constructed a few-shot image segmentation mechanism using a deep convolutional network trained episodically and developed an efficient global correlation module to model the correlation between a support and query image and incorporate it into the deep network.
- Improved Distribution Matching for Dataset Condensation189
This paper identifies two important shortcomings of naive distribution matching and addresses them with three novel techniques and outperforms most previous optimization-oriented methods with much fewer computational resources, thereby scaling data condensation to larger datasets and models.
- 188
- Multi-Source Weak Supervision for Saliency Detection187
A unified framework to train saliency detection models with diverse weak supervision sources is proposed, using category labels, captions, and unlabelled data for training, yet other supervision sources can also be plugged into this flexible framework.
- LSTM-CF: Unifying Context Modeling and Fusion with LSTMs for RGB-D Scene Labeling172
This paper addresses the problem of semantic labeling of RGB-D scenes by developing a novel Long Short-Term Memorized Context Fusion (LSTM-CF) Model that captures and fuses contextual information from multiple channels of photometric and depth data, and incorporating this model into deep convolutional neural networks (CNNs) for end-to-end training.
- Cross-Domain Adaptive Clustering for Semi-Supervised Domain Adaptation171
To achieve both inter-domain and intra-domain adaptation, an adversarial adaptive clustering loss is introduced to group features of unlabeled target data into clusters and perform cluster-wise feature alignment across the source and target domains.
- Data-driven image color theme enhancement170
This paper presents a data-driven method for enhancing a desired color theme in an image that incorporates prior knowledge, such as texture-color relationships, extracted from a database of photographs to maintain a natural look of the edited images.
- Video metamorphosis using dense flow fields: Research Articles167
A novel framework for synthesizing new motions based on given motion manuals and corresponding motion capture examples is presented and it is shown that the smooth concatenation can be achieved by finding an appropriate path in the proposed multi-dimensional index space.
- Inviscid and incompressible fluid simulation on triangle meshes: Research Articles167
A novel framework for synthesizing new motions based on given motion manuals and corresponding motion capture examples is presented and it is shown that the smooth concatenation can be achieved by finding an appropriate path in the proposed multi-dimensional index space.
- Advances in Deep Learning-Based Medical Image Analysis165
Overall, according to the best available evidence, deep learning models performed well in medical image analysis, but what cannot be ignored are the algorithms derived from small-scale medical datasets impeding the clinical applicability.
- Generalized radiograph representation learning via cross-supervision between images and free-text radiology reports164
A method of weak supervision which uses the information of radiology reports to learn visual features without the need for expert labelling is proposed, which has the potential to replace canonical pre-training methodologies.
- TransXNet: Learning Both Global and Local Dynamics With a Dual Dynamic Token Mixer for Visual Recognition148
D-Mixer is proposed, a lightweight dual dynamic token mixer (D-Mixer) to simultaneously learn global and local dynamics, that is, mechanisms that compute weights for aggregating global contexts and local details in an input-dependent manner.
- 147
- Lower Circulating Interferon-Gamma Is a Risk Factor for Lung Fibrosis in COVID-19 Patients143
Data indicate that decreased circulating IFN-γ is a risk factor of lung fibrosis in COVID-19, and comparable with c-reactive protein in the discrimination of the occurrence of lung Fibrotic volume among patients at discharge.
- 143
- GraphFPN: Graph Feature Pyramid Network for Object Detection141
This paper proposes graph feature pyramid networks that are capable of adapting their topological structures to varying intrinsic image structures, and supporting simultaneous feature interactions across all scales, and introduces two types of local channel attention for graph neural networks by generalizing global channel Attention for convolutional neural networks.
- DeepSketch2Face138
A deep learning based sketching system for 3D face and caricature modeling that allows the user to draw freehand imprecise yet expressive 2D lines representing the contours of facial features and a significantly expanded face database with diverse identities, expressions and levels of exaggeration is constructed.
- Harvesting Discriminative Meta Objects with Deep CNN Features for Scene Classification136
This paper presents a novel pipeline built upon deep CNN features to harvest discriminative visual objects and parts for scene classification, and confirms that the scene image representation obtained is capable of delivering state-of-the-art performance on two popular scene benchmark datasets.
- The long-term health outcomes, pathophysiological mechanisms and multidisciplinary management of long COVID135
This work comprehensively review the current literature on long COVID, highlighting its epidemiological understanding, the impact of vaccinations, organ-specific sequelae, pathophysiological mechanisms, and multidisciplinary management strategies, as well as discussing the corresponding recommendations.
- OverLoCK: An Overview-first-Look-Closely-next ConvNet with Context-Mixing Dynamic Kernels134
OverLoCK is presented, the first pure ConvNet backbone architecture that explicitly incorporates a top-down attention mechanism, and a novel context-mixing dynamic convolution (ContMix) is proposed that effectively models long-range dependencies while preserving inherent local inductive biases even when the input resolution increases, addressing critical limitations in existing convolutions.
- Out-of-core tensor approximation of multi-dimensional matrices of visual data131
Experimental results indicate that the techniques in this paper can not only process out-of-core data, but also achieve higher compression ratios and quality than previous methods.
- Swin-UMamba†: Adapting Mamba-Based Vision Foundation Models for Medical Image Segmentation126
Swin-UMamba† is introduced, a novel Mamba-based model for medical image segmentation that seamlessly leverages the power of the vision foundation model, which is also computationally efficient with the linear complexity of Mamba.
- Stroke Controllable Fast Style Transfer with Adaptive Receptive Fields124
This paper presents a stroke controllable style transfer network that incorporates different stroke sizes into one single model and demonstrates that with almost the same number of parameters as the previous Fast Style Transfer algorithm, this network can transfer an artistic style in a stroke controlled manner.
- Graph-Structured Referring Expression Reasoning in the Wild119
A scene graph guided modular network (SGMN), which performs reasoning over a semantic graph and a scene graph with neural modules under the guidance of the linguistic structure of the expression, which significantly outperforms existing state-of-the-art algorithms on the new Ref-Reasoning dataset.
- Single-view hair modeling for portrait manipulation119
A single-view hair modeling technique with modest user interaction is developed to meet the unique requirements set by portrait manipulation and enables a number of interesting applications that were previously challenging, including transferring the hairstyle of one subject to another in a potentially different pose.
- 115
- Preservational Learning Improves Self-supervised Medical Image Models by Reconstructing Diverse Contexts112
This work introduces Preservational Learning to reconstruct diverse image contexts in order to preserve more information in learned representations and provides very competitive results under the pretraining-finetuning protocol, outperforming both self-supervised and supervised counterparts in 5 classification/segmentation tasks substantially.
- Advancing Radiograph Representation Learning with Masked Record Modeling110
It is found that MRM offers superior performance in label-efficient fine-tuning and surpasses self- and report-supervised pre-training in identifying the pneumonia type and the pneumothorax area, sometimes by large margins.
- Cross-Modal Self-Attention with Multi-Task Pre-Training for Medical Visual Question Answering110
This work reformulates image feature pre-training as a multi-task learning paradigm and witness its extraordinary superiority, forcing it to take into account the applicability of features for the specific image comprehension task.
- Multicenter cohort study demonstrates more consolidation in upper lungs on initial CT increases the risk of adverse clinical outcome in COVID-19 patients106
There was an association of older age and larger consolidation in upper lungs on admission with higher odds of poor outcomes in patients with COVID-19 pneumonia, and Multivariable logistic regression models were used to identify the risk factors.
- 106
- RegionGPT: Towards Region Understanding Vision Language Model105
It is demonstrated that a universal RGPT model can be effectively applied and significantly enhancing performance across a range of region-level tasks, including but not limited to complex region descriptions, reasoning, object classification, and referring expressions comprehension.
- Domain generalization on medical imaging classification using episodic training with task augmentation105
This paper proposes a novel DG scheme of episodic training with task augmentation to enhance the variety during training task generation to alleviate the risk of overfitting to training tasks in the paradigm.
- Example-based image color and tone style enhancement104
The goal in this paper is to learn implicit color and tone adjustment rules from examples, and to approximate complicated spatially varying nonlinear mappings in a piecewise manner.
- 103
- SegMAN: Omni-scale Context Modeling with State Space Models and Local Attention for Semantic Segmentation100
This paper introduces SegMAN, a novel linear-time model comprising a hybrid feature encoder dubbed SegMAN Encoder and a decoder based on state space models, enabling highly efficient global context modeling while preserving fine-grained local details.
- 92
- Contrast-Oriented Deep Neural Networks for Salient Object Detection92
This paper develops hybrid contrast-oriented deep neural networks that can significantly outperform the state of the art in terms of all popular evaluation metrics.
- 90
- A Unified Visual Information Preservation Framework for Self-supervised Pre-training in Medical Image Analysis87
The proposed unified SSL framework (PCRLv2) surpasses its self-supervised counterparts on various tasks, including brain tumor segmentation (BraTS 2018), chest pathology identification (ChestX-ray, CheXpert), pulmonary nodule detection (LUNA), and abdominal organ segmentsation (LiTS), sometimes outperforming them by large margins with limited annotations.
- MVP-Net: Multi-view FPN with Position-Aware Attention for Deep Universal Lesion Detection83
With the proposed model design, the data-hunger problem is relieved as the learning task is made easier with the correctly induced clinical practice prior, and promising results are shown on the NIH DeepLesion dataset.
- 83
- I 3 Net: Implicit Instance-Invariant Network for Adapting One-Stage Object Detectors80
An Implicit Instance-Invariant Network (I3Net), which is tailored for adapting one-stage detectors and implicitly learns instance-invariant features via exploiting the natural characteristics of deep features in different layers and exceeds the state-of-the-art performance on benchmark datasets.
- A deformation transformer for real-time cloth animation80
A hybrid method for real-time cloth animation that relies on data-driven models to capture the relationship between cloth deformations at two resolutions and develops a fast collision detection and handling scheme based on dynamically transformed bounding volumes.
- VLN-R1: Vision-Language Navigation via Reinforcement Fine-Tuning79
VLN-R1 proves LVLMs can drive embodied navigation and enhance task-specific reasoning through data-efficient, reward-driven post-training.
- Fast and exact discrete geodesic computation based on triangle-oriented wavefront propagation77
A new geodesic computation algorithm based on a triangle-oriented region growing scheme that can remove most of the redundant windows at the earliest possible stage, thus significantly reducing computational cost and memory usage at later stages.
- Coarse-to-Fine Domain Adaptive Semantic Segmentation with Photometric Alignment and Category-Center Regularization76
This paper proposes a novel UDA pipeline that unifies image-level alignment and category-level feature distribution regularization in a coarse-to-fine manner and improves the generalization capability of the final segmentation model and significantly outperforms all previous state-of-the-arts models.
- Relationship-Embedded Representation Learning for Grounding Referring Expressions76
A CMRE to adaptively highlight objects and relationships related to the given expression with a cross-modal attention mechanism, and represent the extracted information as a language-guided visual relation graph and a Gated Graph Convolutional Network to compute multimodal semantic contexts.
- ConvNets vs. Transformers: Whose Visual Representations are More Transferable?75
This work systematically investigates the transfer learning ability of ConvNets and vision transformers in 15 single-task and multi-task performance evaluations and finds that two ViT models heavily rely on whole network fine-tuning to achieve performance gains while Swin Transformer does not have such a requirement.
- Predicting EGFR and PD-L1 Status in NSCLC Patients Using Multitask AI System Based on CT Images74
Both clinical indicators and radiomics features showed a complementary role in prediction and provided accurate estimates to predict EGFR and PD-L1 status and this non-invasive, high-throughput, and interpretable AI system can be used as an assistive tool in conjunction with or in lieu of ancillary tests and extensive diagnostic workups.
- Compound Domain Generalization via Meta-Knowledge Encoding73
This paper presents COmpound domain generalization via Meta-knowledge ENcoding (COMEN), a general approach to automatically discover and model latent domains in two steps, and introduces Style-induced Domain-specific Normalization (SDNorm) to re-normalize the multi-modal underlying distributions.
- Non-Invasive Measurement Using Deep Learning Algorithm Based on Multi-Source Features Fusion to Predict PD-L1 Expression and Survival in NSCLC73
It is indicated that deep learning model combined with clinical characteristics improved prediction capabilities, which would assist physicians in making rapid decision on clinical treatment options.
- CarveMix: A Simple Data Augmentation Method for Brain Lesion Segmentation73
This work proposes a simple yet effective data augmentation approach, dubbed as CarveMix, for CNN-based brain lesion segmentation, and proposes to model the mass effect that is unique to whole brain tumor segmentation during image mixing.
- 73
- A Structure-Aware Relation Network for Thoracic Diseases Detection and Segmentation72
Experimental results show that the SAR-Net can enhance the strong baseline of Mask R-CNN with significant improvements, and towards making a practical system, the ChestX-Det, a chest X-Ray dataset with instance-level annotations, is provided.
- Attribute Surrogates Learning and Spectral Tokens Pooling in Transformers for Few-shot Learning71
This paper presents new hierarchically cascaded transformers that can improve data efficiency through attribute surrogates learning and spectral tokens pooling and shows clear advantages over SOTA few-shot classification methods in both 5- way 1-shot and 5-way 5-shot settings on four popular benchmark datasets.
- Scene-Intuitive Agent for Remote Embodied Visual Grounding71
This paper focuses on the Remote Embodied Visual Referring Expression in Real Indoor Environments task, called REVERIE, where an agent is asked to correctly localize a remote target object specified by a concise high-level natural language instruction, and proposes a two-stage training pipeline.
- 69
- 69
- ReCoNet: Real-Time Coherent Video Style Transfer Network68
A novel real-time video style transfer model, ReCoNet, is proposed, which can generate temporally coherent style transfer videos while maintaining favorable perceptual styles and exhibits outstanding performance both qualitatively and quantitatively.
- Non-Local Context Encoder: Robust Biomedical Image Segmentation against Adversarial Attacks68
Non-local context encoder (NLCE) is proposed to model short- and long-range spatial dependencies and encode global contexts for strengthening feature activations by channel-wise attention and improves the robustness of other CNN-based biomedical image segmentation methods.
- 67
- A deep-learning model for intracranial aneurysm detection on CT angiography images in China: a stepwise, multicentre, early-stage clinical validation study64
This AI model demonstrated strong clinical potential for intracranial aneurysm detection with improved clinician diagnostic performance, high acceptance, and practical implementation in real-world clinical cases.
- Graph Convolution Based Cross-Network Multiscale Feature Fusion for Deep Vessel Segmentation63
A novel hybrid deep neural network designed to tackle the challenges caused by the sparsity and anisotropy of vessels achieves state-of-the-art 3D vessel segmentation performance on multiple public and in-house datasets.
- A multicenter clinical AI system study for detection and diagnosis of focal liver lesions62
The evidence suggests that LiAIDS can serve as a routine diagnostic tool and enhance the diagnostic capabilities of radiologists for liver lesions, and with the assistance of LiAIDS, the diagnostic accuracy of all radiologists improved.
- Relation Matters: Foreground-aware Graph-based Relational Reasoning for Domain Adaptive Object Detection62
A new and general framework for DAOD is proposed, named Foreground-aware Graph-based Relational Reasoning (FGRR), which incorporates graph structures into the detection pipeline to explicitly model the intra- and inter-domain foreground object relations on both pixel and semantic spaces, thereby endowing the DAOD model with the capability of relational reasoning beyond the popular alignment-based paradigm.
- SSMD: Semi-Supervised Medical Image Detection with Adaptive Consistency and Heterogeneous Perturbation62
The motivation behind SSMD is to provide free yet effective supervision for unlabeled data, by regularizing the predictions at each position to be consistent, and a novel adaptive consistency cost function is developed to regularize different components in the predictions.
- 61
- Interactive Image Segmentation Based on Level Sets of Probabilities60
A robust and accurate algorithm for interactive image segmentation that improves the performance of both the probabilistic classifier and the level set method over multiple passes and makes the final object segmentation less sensitive to user interactions.
- Hierarchical deep network with uncertainty-aware semi-supervised learning for vessel segmentation59
A hierarchical deep network where an attention mechanism localizes the low-contrast capillary regions guided by the whole vessels, and enhance the spatial activation in those areas for the sub-type vessels to alleviate the annotation-hungry limitation of deep models.
- Diagnose Like a Radiologist: Hybrid Neuro-Probabilistic Reasoning for Attribute-Based Medical Image Diagnosis59
A hybrid neuro-probabilistic reasoning algorithm for verifiable attribute-based medical image diagnosis and the ablation study indicates that the hybrid algorithm achieves a much better generalization performance than a pure neural network architecture under very limited training data.
- M 3 Net: A multi-scale multi-view framework for multi-phase pancreas segmentation based on cross-phase non-local attention58
M3Net is presented, a framework that integrates multi-scale multi-view information for multi-phase pancreas segmentation and two types of non-local attention modules are devised to enhance the high-level feature representation across phases.
- Multi-scale Matching Networks for Semantic Correspondence57
This paper follows the coarse-to-fine matching strategy and build a top-down feature and matching enhancement scheme that is coupled with the multi-scale hierarchy of deep convolutional neural networks that achieves state-of-the-art performance on three popular benchmarks with high computational efficiency.
- Act Like a Radiologist: Towards Reliable Multi-View Correspondence Reasoning for Mammogram Mass Detection57
An anatomy-aware graph convolutional network (AGN), which is tailored for mammogram mass detection and endows existing detection methods with multi-view reasoning ability, is proposed and experiments reveal that AGN significantly exceeds the state-of-the-art performance.
- 56
- Inviscid and incompressible fluid simulation on triangle meshes56
A novel method for inviscid fluid simulation over meshes that can enforce incompressibility on closed surfaces by utilizing a discrete vector field decomposition algorithm and eliminates parametrization distortions is introduced.
- A multimodal integration pipeline for accurate diagnosis, pathogen identification, and prognosis prediction of pulmonary infections55
A multimodal integration (MMI) pipeline to differentiate among bacterial, fungal, and viral pneumonia and pulmonary tuberculosis based on a real-world dataset of 24,107 patients was successfully developed and provided an evident advantage in predicting risks of developing critical illness.
- Refer-it-in-RGBD: A Bottom-up Approach for 3D Visual Grounding in RGBD Images55
This work presents a novel task of 3D visual grounding in single-view RGBD image where the referred objects are often only partially scanned due to occlusion and proposes a bottom-up approach to gradually aggregate content-aware information, effectively addressing the challenge posed by the partial geometry.
- Dynamic evolution of COVID-19 on chest computed tomography: experience from Jiangsu Province of China54
Volume, density, and location of the pulmonary opacity and their evolution on CT varied with disease severity in COVID-19, valuable in understanding the nature of the disease and monitoring the patient’s condition during the course of illness.
- 53
- Context-Aware Semantic Inpainting52
An improved GAN-based framework consists of a fully convolutional design for the generator which helps to better preserve spatial structures and a joint loss function with a revised perceptual loss to capture high-level semantics in the context.
- Person Re-Identification Using Multiple Experts with Random Subspaces52
This paper presents a simple and effective multi- expert approach based on random subspaces for person re- identification across non-overlapping camera views that is able to significantly improve the performance of existing state-of-the-art techniques.
- Magic decorator52
The Magic Decorator system is presented, a system that automatically generates material suggestions for 3D indoor scenes and can automatically and efficiently produce a series of visually plausible material suggestions which are comparable to those produced by artists.
- Predicting Invasiveness of Lung Adenocarcinoma at Chest CT with Deep Learning Ternary Classification Models50
Combining framework optimization, joint learning, and an adjudication approach improved DL classification of adenocarcinoma invasiveness at chest CT and improved DL model performance in external testing.
- 50
- A Causal Inference Look at Unsupervised Video Anomaly Detection46
A causal graph is proposed to analyze the confounding effect of the pseudo label generation process and a simple yet effective causal inference based framework is introduced to disentangle the noisy pseudo label's impact to make final anomaly detection.
- AI-based analysis of CT images for rapid triage of COVID-19 patients46
Both clinical and image features showed complementary roles in prediction and provided accurate estimates to the time of progression and the findings are valuable for optimizing the use of medical resources in the COVID-19 pandemic.
- Automated detection and classification of acute vertebral body fractures using a convolutional neural network on computed tomography45
A multistage deep learning system (multistage AO system) can automatically detect and classify acute vertebral body fractures in the thoracolumbar spine on CT images according to AO classification with high accuracy.
- 44
- 44
- 44
- Globally Guided Progressive Fusion Network for 3D Pancreas Segmentation43
A novel end-to-end network is proposed, Globally Guided Progressive Fusion Network, as an effective and efficient solution to volumetric segmentation, which involves both global features and complicated 3D geometric information.
- CaricatureShop: Personalized and Photorealistic Caricature Sketching43
This paper proposes the first sketching system for interactively personalized and photorealistic face caricaturing, which allows outputting different exaggerations when applying the same sketching on different input figures in term of their different geometric characteristics.
- Scale-Equivalent Distillation for Semi-Supervised Object Detection42
Scale-Equivalent Distillation (SED), which is a simple yet effective end-to-end knowledge distillation framework robust to large object size variance and class imbalance, is introduced.
- ROSA: Robust Salient Object Detection Against Adversarial Attacks42
This paper is the first one that mounts successful adversarial attacks on salient object detection models and verifies that adversarial samples are effective on a wide range of existing methods and proposes a novel end-to-end trainable framework to enhance the robustness for arbitrary FCN-based salient object Detection models against adversarial attack.
- 41
- Transformer guided progressive fusion network for 3D pancreas and pancreatic mass segmentation41
A transformer guided progressive fusion network (TGPFN) that utilizes the global representation captured by the transformer to supplement long-range dependencies lost by convolution operations at different resolutions to alleviate limitation of convolution operation.
- Radiogenomic System for Non-Invasive Identification of Multiple Actionable Mutations and PD-L1 Expression in Non-Small Cell Lung Cancer Based on CT Images41
This study is the first deep learning system to simultaneously analyze 10 molecular expressions, which might be utilized as an assistive tool in conjunction with or in lieu of ancillary testing to support precision treatment options.
- Self-supervised Correction Learning for Semi-supervised Biomedical Image Segmentation41
This work adopts a coarse-to-fine strategy and proposes a self-supervised correction learning paradigm for semi- supervised biomedical image segmentation, and designs a dual-task network, including a shared encoder and two independent decoders for segmentation and lesion region inpainting.
- SparX: A Sparse Cross-Layer Connection Mechanism for Hierarchical Vision Mamba and Transformer Networks40
A new sparse cross-layer connection mechanism termed SparX is proposed to effectively improve cross-layer feature interaction and reuse and design a new family of vision backbone networks with sparsely cross-connected layers, achieving an excellent trade-off among model size, computational cost, memory cost, and accuracy in comparison to its counterparts.
- MASS: Modality-collaborative semi-supervised segmentation by exploiting cross-modal consistency from unpaired CT and MRI images40
This paper proposes Modality-collAborative Semi-Supervised segmentation (i.e., MASS), which utilizes the modality-independent knowledge learned from unpaired CT and MRI scans to regularize deep segmentation models in aspects of both semantic and anatomical spaces.
- Real-time automatic polyp detection in colonoscopy using feature enhancement module and spatiotemporal similarity correlation unit39
This research presents a novel and scalable approaches to integrate artificial intelligence into the clinical practice of Gastroenterology and Hepatology and shows real-time decisions about which treatments to recommend and when to treat patients with Digestive Disease.
- 39
- 39
- A Causal Debiasing Framework for Unsupervised Salient Object Detection38
This paper uses causal intervention to perform de-confounded model training to minimize the contrast distribution bias and proposes an image-level weighting strategy that softly weights each image's importance according to the spatial distribution bias map.
- Development and validation of an abnormality-derived deep-learning diagnostic system for major respiratory diseases38
A preliminary version of DeepMRD TR was deployed into the clinical workflow, which was performed on par with senior experts in disease diagnosis, and demonstrated the potential to accelerate the medical workflow to facilitate early diagnosis as a triage tool for respiratory diseases which supports improved clinical diagnoses and decision-making.
- Propagating Over Phrase Relations for One-Stage Visual Grounding38
A linguistic structure guided propagation network for one-stage phrase grounding that explicitly explores the linguistic structure of the sentence and performs relational propagation among noun phrases under the guidance of the linguistic relations between them.
- ME-PCN: Point Completion Conditioned on Mask Emptiness37
This work presents MEPCN, a point completion network that leverages emptiness in 3D shape space and encodes both the occupied point cloud and the neighboring ‘empty points’, and proves that the ‘emptiness’ design is lightweight and easy to embed in existing methods.
- 37
- 37
- 36
- Inter-domain mixup for semi-supervised domain adaptation36
This paper introduces a cross-domain feature alignment strategy, Inter-domain Mixup, that incorporates label information into model adaptation and outperforms existing state-of-the-art methods, achieving significant accuracy improvements on popular SSDA benchmarks, including DomainNet, Office-Home, and Office-31.
- 36
- Piecewise Flat Embedding for Image Segmentation36
A new multi-dimensional nonlinear embedding—Piecewise Flat Embedding (PFE)—for image segmentation based on the theory of sparse signal recovery, which offers an image representation with higher region identifiability which is desirable for image segmentsation or high-level semantic analysis tasks.
- Depthwise Nonlocal Module for Fast Salient Object Detection Using a Single Thread35
A novel depthwise nonlocal module (DNL) is proposed, which implicitly models contrast via harvesting intrachannel and interchannel correlations in a self-attention manner and achieves competitive accuracy and high inference efficiency simultaneously with a single CPU thread.
- Facial Landmark Machines: A Backbone-Branches Architecture With Progressive Representation Learning35
This paper proposes a novel cascaded backbone-branches fully convolutional neural network (BB-FCN) for rapidly and accurately localizing facial landmarks in unconstrained and cluttered settings and confirms that high-quality facial landmarks localized with the proposed network can also improve the precision and recall of face detection.
- Learning Hybrid Representations for Automatic 3D Vessel Centerline Extraction33
This work proposes a hybrid representation learning approach to address the challenge of preserving the continuity of extracted vessels when segmenting such thin tubular structures from 3D images, and validates the proposed approach on CTA datasets and demonstrates its superior performance compared to both traditional and CNN-based baselines.
- Vector solid textures33
A compact random-access vector representation for solid textures made of intermixed regions with relatively smooth internal color variations that is feature-preserving and resolution-independent and enables efficient color evaluation during real-time solid texture mapping.
- DreamDA: Generative Data Augmentation with Diffusion Models32
A new classification-oriented framework DreamDA is proposed, which enables data synthesis and label generation by way of diffusion models and introduces a self-training paradigm for generating pseudo labels and training classifiers using the synthesized data.
- Standard Echocardiographic View Recognition in Diagnosis of Congenital Heart Defects in Children Using Deep Learning Based on Knowledge Distillation32
This study provides a solid foundation for the subsequent use of artificial intelligence (AI) to identify CHDs in children using convolutional neural networks using deep learning-based neural network methods.
- 31
- From Unilateral to Bilateral Learning: Detecting Mammogram Masses with Contrasted Bilateral Network31
This paper proposes a deep model called contrasted bilateral network (CBN) to take bilateral information into consideration, and demonstrates that the proposed algorithm achieves the state-of-the-art performance.
- 31
- I2F: A Unified Image-to-Feature Approach for Domain Adaptive Semantic Segmentation30
A novel UDA pipeline for semantic segmentation that unifies image-level and feature-level adaptation jointly is proposed and Experimental results demonstrate that the pipeline significantly outperforms previous methods.
- Generator Versus Segmentor: Pseudo-healthy Synthesis30
Qualitative and quantitative experimental results on public datasets BraTS and LiTS demonstrate that the proposed method outperforms state-of-the-art methods by preserving style and removing the artifacts.
- The Feature Ambiguity Mitigate Operator model helps improve bone fracture detection on X-ray radiograph29
In bone fracture detection, FAMO is an effective preprocessor to enhance model performance by mitigating feature ambiguity in the network.
- Improving the Diagnostic Accuracy of Breast BI-RADS 4 Microcalcification-Only Lesions Using Contrast-Enhanced Mammography29
CEM has the potential to aid in the diagnosis of BI-RADS 4 calcification-only lesions; in particular, those presented as low risk in morphology and/or distribution may benefit more.
- Less Could Be Better: Parameter-efficient Fine-tuning Advances Medical Vision Foundation Models28
This study delved into LoRA, a representative PEFT method, and compared it against full-parameter fine-tuning (FFT) on two self-supervised radiography foundation models across three well-established chest radiograph datasets, showing that LoRA outperformed FFT in 13 out of 18 transfer learning tasks.
- 28
- 27
- 27
- A fast modal space transform for robust nonrigid shape retrieval27
The approach constructs a context-aware integral kernel operator on a manifold, then applies modal analysis to map this operator into a low-frequency functional representation, called fast functional transform, and finally computes its spectrum as the shape signature.
- 27
- Detection of Intracranial Aneurysms Using Multiphase CT Angiography with a Deep Learning Model26
A multiphase fusion deep learning model with automatic phase selection provided automated detection of IAs with high sensitivity and was compared to the single-phase algorithm to evaluate algorithm sensitivity.
- 26
- OVER-NAV: Elevating Iterative Vision-and-Language Navigation with Open-Vocabulary Detection and StructurEd Representation25
Over-NAV introduces a structured representation, coded Omnigraph, to effectively integrate multi-modal information along the tour and introduces reliable cross-modal supervision and enables on-the-fly generalization to unseen scenes without the need of extra annotation and re-training.
- A multi-stage ensemble network system to diagnose adolescent idiopathic scoliosis25
A deep learning algorithm to automatically evaluate and diagnose scoliosis on full spinal X-ray images collected from four hospital databases had high sensitivity and accuracy forScoliosis, which could help radiologists improve their diagnostic efficiency.
- Revisiting 3D Context Modeling with Supervised Pre-training for Universal Lesion Detection in CT Slices25
A Modified Pseudo-3D Feature Pyramid Network (MP3D FPN) is proposed that leverages depthwise separable convolutional filters and a group transform module (GTM) to efficiently extract 3D context enhanced 2D features for universal lesion detection in CT slices.
- 25
- Harnessing 2D Networks and 3D Features for Automated Pancreas Segmentation from Volumetric CT Images25
In the 3D coarse segmentation network, a new bias-dice loss function is defined to increase ROI recall rates to improve efficiency by avoiding iterative ROI refinements and a fusion decision module and parallel training strategy are proposed to fuse multi-source feature cues extracted from three sub-networks to make final predictions.
- audeosynth25
A framework for automatically generating music-driven video montages is developed that extracts carefully designed temporal features from the input, and casts the synthesis problem as an optimization and solves the parameters through Markov Chain Monte Carlo sampling.
- Fast nonrigid 3D retrieval using modal space transform24
This paper constructs a context-aware integral kernel operator on a manifold, then applies modal analysis to map this operator into a low-frequency functional representation, called fast functional transform, and finally computes its spectrum as the shape signature.
- Artificial Intelligence–Assisted Bone Age Assessment to Improve the Accuracy and Consistency of Physicians With Different Levels of Experience23
Information from an artificial intelligence (AI) deep learning convolutional neural network model improves both the accuracy and the consistency of bone age assessments for physicians of all levels of experience.
- EGC: Image Generation and Classification via a Diffusion Energy-Based Model22
This work introduces an energy-based classifier and generator, namely EGC, which can achieve superior performance in both tasks using a single neural network, and believes that EGC bridges the gap between discriminative and generative learning.
- Advancing 3D medical image analysis with variable dimension transform based supervised 3D pre-training22
An innovative yet simple fully-supervised 3D network pre-training framework is revisited to take advantage of semantic supervisions from large-scale 2D natural image datasets to address the problem of data scarcity and develop powerful 3D representations.
- Adaptive Betweenness Clustering for Semi-Supervised Domain Adaptation22
This study proposes a novel SSDA approach named Graph-based Adaptive Betweenness Clustering (G-ABC) for achieving categorical domain alignment, which enables cross-domain semantic alignment by mandating semantic transfer from labeled data of both the source and target domains to unlabeled target samples.
- 22
- Bilateral Asymmetry Guided Counterfactual Generating Network for Mammogram Classification21
A new theoretical result for counterfactual generation based on the symmetric prior for bilateral images is derived and the utility of the method and the effectiveness of each module in the model can be verified by state-of-the-art performance on INBreast and an in-house dataset and ablation studies.
- 21
- 21
- Automatic Detection of Secundum Atrial Septal Defect in Children Based on Color Doppler Echocardiographic Images Using Convolutional Neural Networks20
The proposed system can automatically and accurately identify ASD, laying a good foundation for the subsequent artificial intelligence diagnosis of CHDs.
- RGB-D Scene Labeling with Long Short-Term Memorized Fusion Model.20
This paper addresses the problem of semantic labeling of RGB-D scenes by developing a novel Long Short-Term Memorized Fusion (LSTM-F) Model that captures and fuses contextual information from multiple channels of photometric and depth data, and incorporating this model into deep convolutional neural networks (CNNs) for end-to-end training.
- Protein Representation Learning via Knowledge Enhanced Primary Structure Modeling19
The proposed Knowledge-exploited Auto-encoder for Protein (KeAP), which performs tokenlevel knowledge graph exploration for protein representation learning, and can consistently outperform the previous counterpart on 9 representative downstream applications, sometimes surpassing it by large margins.
- The application of artificial intelligence to chest medical image analysis19
Four key components of deep learning were summarized, namely, network architectures, learning strategies, optimization methods, and vision tasks, and five promising research directions and possible solutions were presented for future work.
- Contralaterally Enhanced Networks for Thoracic Disease Detection18
A deep end-to-end module to exploit the contralateral context information for enhancing feature representations of disease proposals is proposed, which achieves state-of-the-art performance in weakly-supervised disease localization.
- Comparison of automated and manual DWI-ASPECTS in acute ischemic stroke: total and region-specific assessment18
The eDWI-ASPECTS performed equally well as senior neuroradiologists’ evaluation, although interference by uncertain scoring rules and midline shift resulted in poor to moderate consistency in the M5, internal capsule, and caudate nucleus regions.
- Activate and Reject: Towards Safe Domain Generalization under Category Shift17
A novel Activate and Reject (ART) framework to reshape the model’s decision boundary to accommodate unknown classes and conduct post hoc modification to further discriminate known and unknown classes using unlabeled test data is proposed and experiments reveal that ART consistently improves the generalization capability of deep networks on different vision tasks.
- Structure-aware Meta-fusion for Image Super-resolution17
Experimental results indicate that the meta fusion network outperforms existing state-of-the-art SISR algorithms on widely used datasets, and is capable of producing high-resolution images that achieve low distortion and high perceptual quality simultaneously.
- 17
- Deep reconstruction-recoding network for unsupervised domain adaptation and multi-center generalization in colonoscopy polyp detection17
A domain adaptation model called Deep Reconstruction-Recoding Network (DRRN) is proposed, which jointly learns a shared encoding representation for two tasks: i) a supervised object detection network for labeled source data, and ii) an unsupervised reconstruction-recoding network for unlabeled target data.
- MVCNet: Multiview Contrastive Network for Unsupervised Representation Learning for 3-D CT Lesions16
It is suggested that contrasting multiple 2-D views is an effective approach to capturing the original 3-D information, which notably improves the utilization of the scarce and valuable annotated CT data.
- Automatic object extraction from images using deep neural networks and the level‐set method16
The authors propose an automatic method for extracting objects with fine quality from photographs by finding bounding boxes that enclose potential objects, which is achievable by state-of-the-art object proposal methods and a new multi-pass level-set method based on saliency detection and foreground pixel classification.
- 15
- WaveNano: a signal‐level nanopore base‐caller via simultaneous prediction of nucleotide labels and move labels through bi‐directional WaveNets15
The signal-level nanopore base-caller WaveNano can obtain higher base-calling accuracy, and generate fewer insertions/deletions in the base-called sequences.
- 14
- Efficient visibility processing for projective texture mapping14
This paper presents a novel visibility algorithm for determining which polygons are visible from each camera position that has object-space precision, but operates in both image-space and object- space to get better performance.
- GREN: Graph-Regularized Embedding Network for Weakly-Supervised Disease Localization in X-Ray Images13
This paper proposes the Graph Regularized Embedding Network (GREN), which leverages the intra-image and inter-image information to locate diseases on chest X-ray images and achieves the state-of-the-art result on NIH chestX-ray dataset for weakly-supervised disease localization.
- Harmonizing Pathological and Normal Pixels for Pseudo-Healthy Synthesis13
A new type of discriminator is presented, the segmentor, to accurately locate the lesions and improve the visual quality of pseudo-healthy images, and a reliable metric is proposed by utilizing two attributes of label noise to measure the health of synthetic images.
- Compare and contrast: Detecting mammographic soft-tissue lesions with C 2 -Net13
A multi-view network for breast soft-tissue lesion detection called C2-Net (Compare and Contrast, C2) that fuses information across different views that has achieved state-of-the-art performance.
- BR-GAN: Bilateral Residual Generating Adversarial Network for Mammogram Classification13
Bilateral residual generating adversarial network (BR-GAN) is proposed, which uses the Cycle-GAN as a basic framework while regarding the contralateral as generation reference based on the bilateral symmetry prior and outperforms current state-of-the-art methods on INBreast and in-house datasets.
- Towards Robust Bone Age Assessment: Rethinking Label Noise and Ambiguity13
This work formulate TW3 BAA as an ordinal regression problem, and address both label noise and ambiguity with a two stage deep learning framework, which outperforms previously reported TW3-based BAA systems by large margins.
- 13
- A heuristic method for rapid and automatic radiofrequency ablation planning of liver tumors12
A heuristic RFA planning method to rapidly and automatically generate clinically acceptable RFA plans with multiple clinical constraints is developed and demonstrates the effectiveness of the proposed method and can help reduce the burden on clinicians.
- Diagnostic performance of a deep learning-based method in differentiating malignant from benign subcentimeter (≤10 mm) solid pulmonary nodules12
This deep learning-based model for differentiating malignant and benign solid pulmonary nodules from benign ones in computed tomography images compared against radiologists with 10 and 15 years of experience in thoracic imaging is compared against.
- A pre-treatment CT-based weighted radiomic approach combined with clinical characteristics to predict durable clinical benefits of immunotherapy in advanced lung cancer12
The adoption of weighted radiomic features from multiple intrapulmonary lesions has the potential to predict long-term PFS benefits for patients who are candidates for PD-1/PD-L1 immunotherapies.
- A deep learning algorithm to improve readers’ interpretation and speed of pancreatic cystic lesions on dual-phase enhanced CT12
The DLM achieved senior radiologist-level performance in differentiating benign and malignant PCLs which could improve the junior radiologist’s interpretation and speed ofPCLs on CT.
- Noise2Grad: Extract Image Noise to Denoise12
The proposed method to form the necessary supervision is to extract the noise from the noisy image to synthesize new data, and uses a noise removal module to aid noise extraction to ease the interference of the image background.
- Clinical Applicable AI System Based on Deep Learning Algorithm for Differentiation of Pulmonary Infectious Disease12
For classifying common types of pneumonia and assessing the influential factors for triage, the AI system has shown promising results and is presented to assist clinicians in making quick and accurate diagnoses, resulting in the potential for early therapeutic intervention.
- 12
- Uncertainty Estimation of Large Language Models in Medical Question Answering11
This work benchmarks popular UE methods with different model sizes on medical question-answering datasets, and proposes Two-phase Verification, a probability-free Uncertainty Estimation approach that achieves the best overall accuracy and stability across various datasets and model sizes.
- A deep learning model for prediction of post hepatectomy liver failure after hemihepatectomy using preoperative contrast-enhanced computed tomography: a retrospective study11
The deep learning model showed excellent performance in prediction of PHLF and could be useful for identifying high-risk patients to modify the treatment planning.
- Radiomics Based on Contrast-Enhanced MRI in Differentiation Between Fat-Poor Angiomyolipoma and Hepatocellular Carcinoma in Noncirrhotic Liver: A Multicenter Analysis11
The proposed radiomics model based on triple-phase CE-MRI images was useful for differentiating between fp-AML and HCC and yielded comparable or better performance than two radiologists in different centers, with different scanners and different scanning parameters.
- Generalized Organ Segmentation by Imitating One-Shot Reasoning Using Anatomical Correlation11
Extensive experiments show that OrganNet can effectively resist the wide variations in organ morphology and produce state-of-the-art results in one-shot segmentation task and even when compared with fully-supervised segmentation models, OrganNet is still able to produce satisfying segmentation results.
- Dependence of radiomic features on pixel size affects the diagnostic performance of radiomic signature for the invasiveness of pulmonary ground-glass nodule11
Reducing pixel size could cause inconsistency in most radiomic features and improve the diagnostic performance of the downstream radiomic signatures for the invasiveness for pulmonary ground-glass nodules (GGNs) on CTs.
- A compact random-access representation for urban modeling and rendering11
A highly memory-efficient representation for modeling and rendering urban buildings composed predominantly of rectangular block structures, which can be used to completely or partially represent most modern buildings.
- Automatic Segmentation and Molecular Subtype Classification of Breast Cancer Using an MRI-based Deep Learning Framework10
The proposed novel deep learning framework based on MRI achieved high, robust performance in fully automatic classification of breast cancer molecular subtypes.
- Intracranial aneurysm instability prediction model based on 4D-Flow MRI and HR-MRI10
The Hybrid model provided a promising prediction of aneurysm instability and was further validated in the validation cohort (patients enrolled between December 2021 to May 2022), which provided a promising prediction of aneurysm instability.
- 10
- 10
- 10
- 10
- Automatic Photo Adjustment Using Deep Learning.10
This paper explains how to formulate the automatic photo adjustment problem in a way suitable for deep learning and introduces an image descriptor that accounts for the local semantics of an image that can model local adjustments that depend on the image semantics.
- Modeling and editing real scenes with image-based techniques10
This thesis focuses on capturing and modeling real scenes for novel visual interactions such as varying lighting condition and scene configuration in addition to viewpoint, which can lead to applications such as virtual navigation of a real scene, interaction with the scene, novel scene composition, interior lighting design, and augmented reality.
- SketchMetaFace: A Learning-Based Sketching Interface for High-Fidelity 3D Character Face Modeling9
SketchMetaFace - a sketching system targeting amateur users to model high-fidelity 3D faces in minutes and a novel learning-based method termed “Implicit and Depth Guided Mesh Modeling” (IDGMM).
- DAE-GCN: Identifying Disease-Related Features for Disease Prediction9
The proposed DAE-GCN helps improve the performance and the inter-pretability of cancer prediction, which can be verified by state-of-the-art performance on one public dataset DDSM and three in-house datasets.
- Segmentation-Based Method Combined with Dynamic Programming for Brain Midline Delineation9
This paper formulate the brain midline delineation as a segmentation task and proposes a three-stage framework that can achieve superior performance on one in-house dataset and one public dataset.
- 9
- 8
- Context-Aware Refinement Network Incorporating Structural Connectivity Prior for Brain Midline Delineation8
A context-aware refinement network (CAR-Net) is proposed to refine and integrate the feature pyramid representation generated by the UNet and explores more discriminative contextual features and larger receptive field, which is of great importance to predict largely deformed midline.
- Simultaneous Lung Field Detection and Segmentation for Pediatric Chest Radiographs8
Experimental results demonstrate that the proposed SDSLung is capable of achieving significantly superior performance over state-of-the-art LFS methods on the authors' large-scale pediatric CXR dataset and also achieving extremely competitive results on adults’ C XR dataset.
- 8
- Parallel Progressive Radiosity with Adaptive Meshing8
An efficient parallel algorithm for progressive radiosity with adaptive meshing is presented, which adopts a novel static processor assignment strategy and takes advantages of hierarchical computation structure in this problem to minimize communication and balance dynamic load.
- Rethinking Discrete Tokens: Treating Them as Conditions for Continuous Autoregressive Image Synthesis7
This work introduces DisCon (DiscreteConditioned Continuous Autoregressive Model), a novel framework that reinterprets discrete tokens as conditional signals rather than generation targets, which circumvents the optimization challenges of continuous token modeling while avoiding the information loss caused by quantization.
- Development and Validation of a Sham-AI Model for Intracranial Aneurysm Detection at CT Angiography7
Radiologists' diagnostic performance was not compromised when aided by the proposed Sham-AI model compared with their unassisted performance, and radiologists' susceptibility to Sham-AI suggestions was assessed.
- Multi-stream Progressive Up-Sampling Network for Dense CT Image Reconstruction7
The improvement in both sensitivity and precision shows the potential value of the proposed method in clinical applications.
- 7
- Action-Gons: Action Recognition with a Discriminative Dictionary of Structured Elements with Varying Granularity7
This paper proposes a systematic learning process to build an over-complete dictionary of “Action-Gons”, a middle level representation for action recognition in videos inspired by the Julesz school of n-gon statistics.
- NavGemini: a multi-modal LLM agent for vision-and-language navigation6
The proposed NavGemini, a navigation system based entirely on the newly developed multi-modal LLM, Gemini-Pro-Vision, successfully outperforms previous methods by 5.7% in terms of success rate, even when using inferior LLMs, demonstrating the strong ability and potential of multi-modal LLM-based agents in VLN tasks.
- Exploration and Exploitation of Unlabeled Data for Open-Set Semi-supervised Learning6
This paper proposes a prototype-based clustering and identification algorithm that explores the inherent similarity and difference among samples at feature level and effectively cluster them around several predefined ID and OOD prototypes, thereby enhancing feature learning and facilitating ID/OOD identification.
- 6
- 6
- A fast propagation scheme for approximate geodesic paths6
A novel framework for efficient and approximate geodesic path computation over triangle meshes is presented, which finds and propagates paths based on a continuous Dijkstra strategy with a two-stage approach to compute a path for each propagating step.
- ZoomTree: Unrestricted Zoom Paths in Multiscale Visual Analysis of Relational Databases6
This work presents a multiscale visualization system supporting unrestricted zoom paths, called “ZoomTree”, and has a flexible visual interface on the client side, and a powerful and efficient back end with GPU-based parallel online data cubing and CPU-based data clustering.
- Object Modeling and Animation with Smoke6
An approach to design an imaginary force field including both long-range and short-range forces as well as a nonuniform biased diffusion equation driven by the target density, which can effectively achieve the desired results.
- ScreenLungNet: Personalized Long-Term Prediction of Lung Cancer Risk from a Single Low-Dose CT Screening5
A model integrating multiple nodules and global lung features outperformed nodule-only models for long-term lung cancer risk prediction and enabled risk prediction in screening-negative participants.
- Autoregressive Sequence Modeling for 3D Medical Image Representation5
A pioneering method for learning 3D medical image representations through an autoregressive pre-training framework that sequences various 3D medical images based on spatial, contrast, and semantic correlations, treating them as interconnected visual tokens within a token sequence.
- Assessing the Impact of an Artificial Intelligence-Based Model for Intracranial Aneurysm Detection in CT Angiography on Patient Diagnosis and Outcomes (IDEAL Study)—a protocol for a multicenter, double-blinded randomized controlled trial5
The precise impact of the AI system on the detection performance for intracranial aneurysms is determined in a double-blinded design and following the real-world effects on patients’ short-term and long-term outcomes.
- 5
- A teacher-student framework for liver and tumor segmentation under mixed supervision from abdominal CT scans5
A hierarchical organ-to-lesion (O2L) attention module is proposed to regularize the teacher annotator trained on the strong dataset and shows robustness to the quality of bounding box annotations.
- Learning Part Generation and Assembly for Sketching Man‐Made Objects5
This paper decouple the task of sketch2shape into a part generation module and a part assembling module, where deep learning methods are leveraged for the implementation of both modules, and eases the learning process of the shape generator and guarantees high‐quality outputs.
- Reliability and Reproducibility of the Diameters Method in Rapid Determination of Acute Infarct Volume in Magnetic Resonance Diffusion-Weighted Imaging5
The MD method was the best for achieving a rapid and excellent interobserver reliability for estimating infarct volume and can quickly screen patients suitable for recanalization treatment and predict poor prognosis through threshold evaluation.
- 5
- Motion Field Estimation for Temporal Textures5
A novel approach based on a local flow probability distribution function at each pixel using the STAR model and the data from a spatio-temporal neighborhood to estimate flow fields of dynamic temporal textures whose motion differs radically from that of rigid bodies.
- A large language model for clinical outcome adjudication from telephone follow-up interviews: a secondary analysis of a multicenter randomized clinical trial4
Fu-LLM effectively automates the preadjudication of key clinical events—including death, hospitalization, and medication use—based on 1,046 vignettes of follow-up telephone interviews conducted across three centers in a randomized clinical trial (China CT-FFR Study 3).
- Leveraging Frequency Domain Learning in 3D Vessel Segmentation4
This study leverages Fourier domain learning as a substitute for multi-scale convo-lutional kernels in 3D hierarchical segmentation models in 3D hierarchical segmentation models to reduce computational expenses while preserving global receptive fields within the network.
- 4
- 4
- 4
- A Rendering Equation for Specular Transfers and Its Integration into Global Illumination4
Intensity and area factors are introduced to clarify this problem and to rectify the radiance from these specular surfaces and an algorithm for integrating ideal specular transfers into global illumination is presented.
- 3
- Multi-Level Contrastive Learning for Dense Prediction Task3
This work presents Multi-Level Contrastive Learning for Dense Prediction Task (MCL), an efficient self-supervised method for learning region-level feature representation for dense prediction tasks, and proposes a novel pretext task that assembles multi-scale images in a montage manner to mimic multi-object scenarios.
- EMS: 3D Eyebrow Modeling from Single-View Images3
This work proposes EMS, the first learning-based framework for single-view 3D eyebrow reconstruction, and builds the first 3D synthetic eyebrow dataset that contains 400 high-quality eyebrow models manually created by artists.
- 3
- 3
- 3
- Laplacian Texture Synthesis and Mixing on Surfaces3
Laplacian texture reconstruction is proposed, which retains the high frequency details but computes new consistent low frequency components to eliminate the seams and can be viewed as a global feature-preserving smoothing step, and is more effective than local feathering.
- Figure 1. A texture metamorphosis sequence is generated from the leftmost to the rightmost. Pattern-based Texture Metamorphosis3
This paper proposes a patternbased approach to specify the feature correspondence between two textures, based on the observation that many texture images have stochastically distributed patterns which are similar to each other.
- A2Mamba: Attention-augmented State Space Models for Visual Recognition2
A2Mamba is proposed, a powerful Transformer-Mamba hybrid network architecture, featuring a new token mixer termed Multi-scale Attention-augmented State Space Model (MASS), where multi-scale attention maps are integrated into an attention-augmented SSM (A2SSM).
- Dynamic Triple Reweighting Network for Automatic Femoral Head Necrosis Diagnosis from Computed Tomography2
This study aims to develop a computer-aided diagnosis system for early AVNFH based on computed tomography (CT) which provides layer-wise features and is less costly, and proposes the Dynamic Triple Reweighting Network (DTRNet), which integrates the AV NFH classification and weakly-supervised localization.
- Domain Invariant Model with Graph Convolutional Network for Mammogram Classification2
A novel framework, namely Domain Invariant Model with Graph Convolutional Network (DIM-GCN), which only exploits invariant disease-related features from multiple domains that are provable to be disentangled from each other.
- 2
- 2
- A robust high-resolution details preserving denoising algorithm for meshes2
This work presents a robust and efficient high-resolution details-preserving mesh denoising algorithm which improves the quality of finely captured meshes and updates the vertex positions to be consistent with the filtered face normals using a least-squares formulation.
- 2
- Deep Learning for Coronary Stenosis Detection in Heavily Calcified Plaques at Coronary CT Angiography: A Stepwise, Multicenter Study1
The proposed DL model accurately detected coronary stenosis of heavily calcified plaques on CCTA and improved diagnostic performance of radiologists and improved diagnostic accuracy among junior, attending, and senior radiologists.
- Visualizing Radiologic Connections: An Explainable Coarse-to-Fine Foundation Model with Multiview Mammograms and Associated Reports1
A foundational pretraining method for digital mammography that extracts fine-grained visual-language representations from images and reports in label-limited settings enhances interpretability and fine-grained multimodal foundational learning for multiview mammograms and reports.
- ViKL: A Mammography Interpretation Framework via Multimodal Aggregation of Visual-knowledge-linguistic Features1
This work proposes ViKL, a innovative framework that synergizes Visual, Knowledge, and Linguistic features that substantially enhances the pathological classification and fosters multimodal interactions in mammography.
- 1
- 1
- 1
- 1
- 1
- 1
- Editorial Special issue on the fifth Computational Visual Media conference (CVM 2017)1
This special issue of Graphical Models includes a selection of oral papers presented at the recent Computational Visual Media conference held at Nankai University, Tianjin, China, and a novel rigidity controllable mesh deformation method, which can flexibly adjust shape rigidity by expanding the local neighborhood around a vertex beyond the first ring.
- A novel method for 3D morphing by deformation matrix with triangle meshes1
A 3D morphing method which generates a merged model given a series of triangle meshes, based on a set of parameters between the source and target shapes, which can show the process of the transformation from the source to the target smoothly.
- 1
- 1
- Linear Subspace Learning based on a Learned Discriminative Dictionary for Sparse Coding1
VISAPP is part of VISIGRAPP - the 9th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications.
- A Robust Algorithm for Denoising Meshes with High-Resolution Details1
A robust and efficient mesh Denoising algorithm which preserves high-resolution details very well and has advantages over previous surface denoising methods.
- Parallel H-Tree Based Data Cubing on Graphics Processors1
This paper investigates efficient GPU-based data cubing because the most frequent operation in data cube computation is aggregation, which is an expensive operation well suited for SIMD parallel processors.
- –
- –
- Latent Diffusion With Topology-Preserving Conditioning for Structure-Aware Data Generation–
Experiments show that TopoSegDiff produces more realistic and structurally faithful image-mask pairs, reduces topology inconsistency, and consistently improves downstream segmentation performance when used for synthetic augmentation, which support topology-preserving conditioning as an effective route toward more reliable structure-aware generation and learning for curvilinear objects.
- –
- –
- –
- Enhanced MRI Representation via Cross-series Masking–
This work introduces a novel Cross-Series Masking (CSM) Strategy for effectively learning MRI representation in a self-supervised manner and achieves state-of-the-art performance on both public and in-house datasets.
- Progressive Conservative Adaptation for Evolving Target Domains–
The proposed PCAda is implemented with a meta-learning framework, which achieves the fast adaptation of the classifier with the help of the progressively updated class prototypes in the inner loop and learns a generalized feature without severely interfering with the historic knowledge via the conservative sparse attention in the outer loop.
- GraVIS: Grouping Augmented Views From Independent Sources for Dermatology Analysis–
GraVIS is proposed, which is specifically optimized for learning self-supervised features from dermatology images, to group homogeneous dermatologyImages while separating heterogeneous ones, and when equipped with the pre-trained weights provided by GraVIS, a single model could achieve better results than winners that heavily rely on ensemble strategies in the well-known ISIC 2017 challenge.
- –
- –
- –
- –
- Harvesting Visual Objects from Internet Images via Deep-Learning-Based Objectness Assessment–
A novel deep convolutional neural network is designed for the inference of proposal objectness, the probability of a proposal containing optimally located foreground object, reflecting two complementary features of an optimal proposal: a complete foreground and relatively small background.
- P038 Investigating the impact of poor sleep on cardiovascular health and cerebrovascular burden in healthy ageing using the UK biobank data–
This exploratory analysis confirms the impact of measures of poor sleep on cerebrovascular health, proposing a complex relationship between sleep and WMH loads involving cardiovascular features in a large ageing population.
- –
- Surface Texture Synthesis and Mixing Using Differential Colors–
Laplacian texture reconstruction is proposed, which retains the high frequency details but computes new consistent low frequency components to eliminate seams and can be viewed as a global feature-preserving smoothing step, and is more effective than local feathering.
- Example-based art pattern synthesis using level sets–
This paper focuses on synthesizing art patterns with curvilinear features from exemplars, which is cast as a global optimization problem and develops an overall EM-style algorithm for minimizing this energy function.
- –
- Medical Volume Segmentation based on Level Sets of Probabilities–
VISAPP is part of VISIGRAPP - the 9th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications.
- MULTISCALE VISUALIZATION OF RELATIONAL DATABASES USING LAYERED ZOOM TREES AND PARTIAL DATA CUBES–
This paper uses zoom trees to represent the entire history of a zooming process that reveals multiscale details and develops efficient GPU-based parallel algorithms for online data cubing and CPU-based data clustering.
- Feature-based texture synthesis and hierarchical tensor approximation–
This work proposes to relax the restrictions and decompose synthesis into two stages: feature map synthesis and Laplacian texture reconstruction, and proves that the hierarchical multilinear models can achieve higher compression ratios and quality on high-dimensional visual data than wavelet (packet) transforms and single-level tensor approximation.
- –
- –
- –
- Efficient and Realistic Hair Modeling and Animation Using Sparse Guide Hairs–
This paper presents a unified framework for modeling and animating curly hairs based on a sparse set of guide strands that can be generated and removed on the fly to dynamically control the accuracy of a simulation.
- –
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.