Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Chang Wen ChenSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works462 from public data
- Recent advances in convolutional neural networks6,348
This paper details the improvements of CNN on different aspects, including layer design, activation function, loss function, regularization, optimization and fast computation, and introduces various applications of convolutional neural networks in computer vision, speech and natural language processing.
- When Gaussian Process Meets Big Data: A Review of Scalable GPs920
This article is devoted to reviewing state-of-the-art scalable GPs involving two main categories: global approximations that distillate the entire data and local approximation that divide the data for subspace learning.
- Auto-Encoding Scene Graphs for Image Captioning857
This work proposes Scene Graph Auto-Encoder (SGAE) that incorporates the language inductive bias into the encoder-decoder image captioning framework for more human-like captions and validates the effectiveness of SGAE on the challenging MS-COCO image captioned benchmark.
- MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-view Images656
MVSplat is introduced, an efficient model that, given sparse multi-view images as input, predicts clean feed-forward 3D Gaussians and builds a cost volume representation via plane sweeping, where the cross-view feature similarities stored in the cost volume can provide valuable geometry cues to the estimation of depth.
- GMFlow: Learning Optical Flow via Global Matching583
A GMFlow framework is proposed, which consists of three main components: a customized Transformer for feature enhancement, a correlation and softmax layer for global feature matching, and a self-attention layer for flow propagation, which outperforms 31-refinements RAFT on the challenging Sintel benchmark, while using only one refinement and running faster.
- 548
- Pluralistic Image Completion539
This paper proposes a novel and probabilistically principled framework with two parallel paths for pluralistic image completion that utilizes the only one given ground truth to get prior distribution of missing parts and rebuild the original image from this distribution.
- 3D Hand Shape and Pose Estimation From a Single RGB Image517
This work proposes a Graph Convolutional Neural Network (Graph CNN) based method to reconstruct a full 3D mesh of hand surface that contains richer information of both 3D hand shape and pose and proposes a weakly-supervised approach by leveraging the depth map as a weak supervision in training.
- Mutual consistency learning for semi-supervised medical image segmentation457
A novel mutual consistency network (MC-Net+) is proposed to effectively exploit the unlabeled data for semi-supervised medical image segmentation to force the model to generate invariant results in such challenging regions, aiming at regularizing the model training.
- A survey of adaptive sampling for global metamodeling in support of simulation-based complex engineering design448
This article categorizes, reviews, and analyzes the state-of-the-art single−/multi-response adaptive sampling approaches for global metamodeling in support of simulation-based engineering design and discusses some important issues that affect the success of an adaptive sampling approach.
- Unifying Flow, Stereo and Depth Estimation433
It is demonstrated that cross-attention enables integration of knowledge from another image via cross-view interactions, which greatly improves the quality of the extracted features and enables cross-task transfer.
- Look, Imagine and Match: Improving Textual-Visual Cross-Modal Retrieval with Generative Models420
This work proposes to incorporate generative processes into the cross-modal feature embedding, through which it is able to learn not only the global abstract features but also the local grounded features of image-text pairs.
- Joint source channel rate-distortion analysis for adaptive mode selection and rate control in wireless video coding386
An analytic solution for adaptive intra mode selection and joint source-channel rate control under time-varying wireless channel conditions is derived and significantly improves the end-to-end video quality in wireless video coding and transmission.
- Semi-supervised Left Atrium Segmentation with Mutual Consistency Training363
A novel Mutual Consistency Network (MC-Net) for semi-supervised left atrium segmentation from 3D MR images that consists of one encoder and two slightly different decoders, and the prediction discrepancies of two decoder are transformed as an unsupervised loss by a designed cycled pseudo label scheme to encourage mutual consistency.
- Scene Graph Generation With External Knowledge and Image Reconstruction330
This paper proposes a novel scene graph generation algorithm with external knowledge and image reconstruction loss to overcome dataset issues, and extracts commonsense knowledge from the external knowledge base to refine object and phrase features for improving generalizability inscene graph generation.
- Fast Vision Transformers with HiLo Attention323
This work introduces LITv2, a simple and effective ViT which performs favourably against the existing state-of-the-art methods across a spectrum of different model sizes with faster speed and shows that HiLo is superior to the existing attention mechanisms by comprehensively benchmarking FLOPs, speed and memory consumption on GPUs and CPUs.
- Weakly-Supervised 3D Hand Pose Estimation from Monocular RGB Images320
A weakly-supervised method, adaptating from fully-annotated synthetic dataset toWeakly-labeled real-world dataset with the aid of a depth regularizer, which generates depth maps from predicted 3D pose and serves as weak supervision for3D pose regression.
- Remarks on multi-output Gaussian process regression313
This article investigates the state-of-the-art multi-output Gaussian processes (MOGPs) that can transfer the knowledge across related outputs in order to improve prediction quality and gives some recommendations regarding the usage of MOGPs.
- Portable Food‐Freshness Prediction Platform Based on Colorimetric Barcode Combinatorics and Deep Convolutional Neural Networks295
Cross‐reactive colorimetric barcode combinatorics and deep convolutional neural networks (DCNNs) are combined to form a system for monitoring meat freshness that concurrently provides scent fingerprint and fingerprint recognition.
- Beyond pixels: A comprehensive survey from bottom-up to semantic image segmentation and cosegmentation240
A comprehensive review of the recent progress in image segmentation, covering 190 publications, gives an overview of broad segmentation topics including not only the classic unsupervised methods, but also the recent weakly-/semi- supervised methods and the fully-super supervised methods.
- 231
- Causal Attention for Vision-Language Tasks228
CATT improves various popular attention-based vision-language models by considerable margins and has great potential in large-scale pre-training, e.g., it can promote the lighter LXMERT, which uses fewer data and less computational power, comparable to the heavier UNITER.
- 217
- T$$^2$$Net: Synthetic-to-Realistic Translation for Solving Single-Image Depth Estimation Tasks204
A framework that comprises an image translation network for enhancing realism of input images, followed by a depth prediction network that can be trained end-to-end, leading to good results, even surpassing early deep-learning methods that use real paired data.
- 199
- Unpaired Image Captioning via Scene Graph Alignments199
This paper proposes an unsupervised feature alignment method that maps the scene graph features from the image to the sentence modality and can generate quite promising results without using any image-caption training pairs, outperforming existing methods by a wide margin.
- 198
- Exploit Bounding Box Annotations for Multi-Label Object Recognition195
This paper first extracts object proposals from each image, then proposes to make use of ground-truth bounding box annotations (strong labels) to add another level of local information by using nearest-neighbor relationships of local regions to form a multi-view pipeline.
- Stack-Captioning: Coarse-to-Fine Learning for Image Captioning194
A coarse-to-fine multi-stage prediction framework for image captioning, composed of multiple decoders each of which operates on the output of the previous stage, producing increasingly refined image descriptions, which simultaneously solves the well-known exposure bias problem and the loss-evaluation mismatch problem.
- Admission control in IEEE 802.11e wireless LANs171
It is shown that the new MAC-layer QoS schemes and parameters provided in EDCA and HCCA can be well utilized to fulfill the requirements of admission control so that QoS for multimedia applications can be provided in WLANs.
- JÂA-Net: Joint Facial Action Unit Detection and Face Alignment Via Adaptive Attention170
A novel end-to-end deep learning framework for joint AU detection and face alignment, which has not been explored before is proposed, in which multi-scale shared feature is learned firstly, and high-level feature of face alignment is fed into AU detection.
- Deconfounded Image Captioning: A Causal Retrospect168
This paper presents a novel perspective: Deconfounded Image Captioning (DIC), to find out the answer of this question, then retrospect modern neural image captioners, and proposes a DIC framework: DICv1.0 to alleviate the negative effects brought by dataset bias.
- Scalable Vision Transformers with Hierarchical Pooling162
A Hierarchical Visual Transformer (HVT) is proposed which progressively pools visual tokens to shrink the sequence length and hence reduces the computational cost, analogous to the feature maps downsampling in Convolutional Neural Networks (CNNs).
- 162
- An Empirical Study of Language CNN for Image Captioning159
This paper introduces a language CNN model which is suitable for statistical language modeling tasks and shows competitive performance in image captioning, and is competitive with the state-of-the-art methods.
- MoVQ: Modulating Quantized Vectors for High-Fidelity Image Generation158
This work proposes to incorporate the spatially conditional normalization to modulate the quantized vectors so as to insert spatially variant information to the embedded index maps, encouraging the decoder to generate more photorealistic images.
- 158
- 154
- 149
- The Spatially-Correlative Loss for Various Image Translation Tasks148
This work proposes a novel spatially-correlative loss that is simple, efficient and yet effective for preserving scene structure consistency while supporting large appearance changes during unpaired image-to-image (I2I) translation, and introduces a new self-supervised learning method to explicitly learn spatially -correlative maps for each specific translation task.
- HMM-Based Audio Keyword Generation148
Experimental results show that, for audio keyword generation, the proposed HMM-based method outperforms the previous hierarchical SVM, and is motivated by the successful story of HMM in speech recognition.
- Learning Object-Language Alignments for Open-Vocabulary Object Detection144
This paper proposes a novel open-vocabulary object detection framework directly learning from image-text pair data as a set matching problem between a set of image region features and aSet of word embeddings, which enables us to train an open- Vocabulary object detector on image- Text pairs in a much simple and effective way.
- 139
- Learning Meta-class Memory for Few-Shot Semantic Segmentation136
This paper proposes a novel Meta-class Memory based few-shot segmentation method (MM-Net), where a set of learnable memory embeddings are introduced to memorize the meta-class information during the base class training and transfer to novel classes during the inference stage.
- 133
- An adaptive sampling approach for Kriging metamodeling by maximizing expected prediction error131
An adaptive sampling approach under the bias-variance decomposition framework that sequentially selects new points by maximizing an expected prediction error criterion that considers both the bias and variance information is proposed.
- VQA-E: Explaining, Elaborating, and Enhancing Your Answers for Visual Questions125
This work proposes a new task of VQA-E (VQA with Explanation), where the computational models are required to generate an explanation with the predicted answer, and quantitatively shows that the additional supervision from explanations can not only produce insightful textual sentences to justify the answers, but also improve the performance of answer prediction.
- RSG: A Simple but Effective Module for Learning Imbalanced Datasets124
This work proposes a new rare-class sample generator (RSG), which aims to generate some new samples for rare classes during training and has in particular the following advantages: it is convenient to use and highly versatile, because it can be easily integrated into any kind of convolutional neural network.
- LLMFormer: Large Language Model for Open-Vocabulary Semantic Segmentation121
This paper introduces a novel paradigm by harnessing the unprecedented capabilities of large language models (LLMs) by capitalizing on the vast knowledge embedded within LLMs for OV semantic segmentation, and proposes three novel attention modules-semantic, scaled visual, and relation attentions to utilize the LLM priors.
- Bridging Global Context Interactions for High-Fidelity Image Completion119
This paper proposes to treat image completion as a directionless sequence-to-sequence prediction task, and deploy a transformer to directly capture long-range depen-dence, using a restrictive CNN with small and non-overlapping RF for weighted token representation.
- CT volumetry of the liver: Where does it stand in clinical practice?119
The various technical factors that contribute to accurate volumetric measurements are explored, and how the accuracies of these techniques are influenced by their methodologies are demonstrated.
- Sensitivity-Aware Visual Parameter-Efficient Fine-Tuning115
A novel Sensitivity-aware visual Parameter-efficient fine-Tuning (SPT) scheme, which adaptively allocates trainable parameters to task-specific important positions given a desired tunable parameter budget, which largely boosts their performance.
- Less Is More: Pay Less Attention in Vision Transformers113
A hierarchical Transformer where pure multi-layer perceptrons (MLPs) are used to encode rich local patterns in the early stages while applying self-attention modules to capture longer dependencies in deeper layers is proposed.
- MARLIN: Masked Autoencoder for facial video Representation LearnINg109
The proposed framework, named MARLIN, is a facial video masked autoencoder that learns highly robust and generic facial embeddings from abundantly available non-annotated web crawled facial videos that can transfer across a variety of facial analysis tasks such as Facial Attribute Recognition (FAR), Facial Expression Recognition, DeepFake Detection, and Lip Synchronization.
- 108
- 102
- 101
- 100
- High-Resolution Optical Flow from 1D Attention and Correlation100
A novel 1D formulation empowers the proposed method to scale to very high-resolution input images while maintaining competitive performance, and to achieve 2D correspondence modeling effect.
- Learning to Collocate Neural Modules for Image Captioning97
This work proposes a novel framework: learning to Collocate Neural Modules (CNM), to generate the ``inner pattern'' connecting visual encoder and language decoder, and achieves a new state-of-the-art 127.9 CIDEr-D on Karpathy split and a single-model 126.0 c40 on the official server.
- Decoupled Spatial Neural Attention for Weakly Supervised Semantic Segmentation96
A decoupled spatial neural attention network for generating pseudo-annotations is proposed, inspired by spatial neural-attention for image captioning, which could simultaneously identify the object regions and localize the discriminative parts, which generates high-quality pseudo- annotations in one forward path.
- Facial Motion Prior Networks for Facial Expression Recognition94
This work proposes a novel FER framework, named Facial Motion Prior Networks (FMPN), which introduces an addition branch to generate a facial mask so as to focus on facial muscle moving regions.
- Unpaired Image Captioning by Language Pivoting92
This work can effectively capture the characteristics of an image captioner from the pivot language (Chinese) and align it to the target language (English) using another pivot-target (Chinese-English) parallel corpus.
- DrVideo: Document Retrieval Based Long Video Understanding89
DrVideo is proposed, a document-retrieval-based system designed for long video understanding that significantly outperforms existing LLM-based state-of-the-art methods on EgoSchema benchmark, MovieChat-1K benchmark, and the long split of Video-MME benchmark.
- QLLM: Accurate and Efficient Low-Bitwidth Quantization for Large Language Models89
QLLM is proposed, an accurate and efficient low-bitwidth PTQ method designed for LLMs that introduces an adaptive channel reassembly technique that reallocates the magnitude of outliers to other channels, thereby mitigating their impact on the quantization range.
- 87
- 87
- Variational mesh decomposition87
A variational mesh decomposition algorithm that can efficiently partition a mesh into a prescribed number of segments is presented that outperforms competitive segmentation methods when evaluated on the Princeton Segmentation Benchmark.
- M2E-Try On Net85
A novel virtual Try-On network, M2E-Try On Net, which transfers the clothes from a model image to a person image without the need of any clean product images is presented.
- 84
- 82
- CSG-Stump: A Learning Friendly CSG-Like Representation for Interpretable Shape Parsing81
CSG-Stump Net is presented, an unsupervised end-to-end network for learning shapes from point clouds and discovering the underlying constituent modeling primitives and operations as well, proven to be equivalent to CSG in terms of representation.
- 81
- ProposalCLIP: Unsupervised Open-Category Object Proposal Generation via Exploiting CLIP Cues79
The ProposalCLIP is able to predict proposals for a large variety of object categories without annotations, by exploiting CLIP (contrastive language-image pre-training) cues, and shows benefits for downstream tasks, such as unsupervised object detection.
- Finding It at Another Side: A Viewpoint-Adapted Matching Encoder for Change Captioning79
A novel visual encoder is proposed to explicitly distinguish viewpoint changes from semantic changes in the change captioning task and further simulate the attention preference of humans and proposes a novel reinforcement learning process to fine-tune the attention directly with language evaluation rewards.
- Cope with diverse data structures in multi-fidelity modeling: A Gaussian process method78
Numerical experiments reveal that this very promising Bayesian MFM approach is capable of effectively extracting the low-f fidelity information for facilitating the modeling of the high-fidelity output using diverse data structures.
- HAC++: Towards 100X Compression of 3D Gaussian Splatting75
This paper proposes HAC++, which explicitly minimizes the 3D Gaussian Splatting representation’s entropy during optimization, enabling efficient arithmetic coding after training for compressed storage and incorporating an adaptive masking strategy to eliminate non-effective Gaussians and anchors.
- Shuffle-Then-Assemble: Learning Object-Agnostic Visual Relationship Features75
Extensive experiments on two visual relationship benchmarks show that by using the novel Shuffle-Then-Assemble pre-trained features, naive relationship models can be consistently improved and even outperform other state-of-the-art relationship models.
- 75
- Unified Open-Vocabulary Dense Visual Prediction73
A Unified Open-Vocabulary Network (UOVN) is proposed to jointly address four common dense prediction tasks and proposes a multi-modal, multi-scale and multi-task (MMM) decoding mechanism to better exploit multi-modal information for OV recognition.
- 73
- MIML-FCN+: Multi-Instance Multi-Label Learning via Fully Convolutional Networks with Privileged Information72
A two-stream fully convolutional network unified by a novel PI loss to solve the problem of MIML learning with privileged bags is proposed, outperforming state-of-the-art methods in the application of multi-object recognition.
- 71
- Pruning Self-Attentions Into Convolutional Layers in Single Path70
This work proposes a novel weight-sharing scheme between MSA and convolutional operations, delivering a single-path space to encode all candidate operations, and introduces learnable binary gates to encode the operation choices in MSA layers.
- 70
- 70
- Improving Multi-label Learning with Missing Labels by Structured Semantic Correlations70
This paper proposes to incorporate structured semantic correlations to solve the missing label problem of multi- label learning and utilizes the semantic graph Laplacian as a smooth term in the multi-label learning formulation to incorporate the structured semantic correlation.
- 69
- Robust surface reconstruction via dictionary learning69
A unified framework that treats geometry and connectivity construction as one joint optimization problem is proposed that outperforms the state-of-the-art in terms of accuracy, robustness to noise and outliers, geometric feature and detail preservation, and mesh connectivity.
- Fast Feedforward 3D Gaussian Splatting Compression67
This work introduces Fast Compression of 3D Gaussian Splatting (FCGS), an optimization-free model that can compress 3DGS representations rapidly in a single feed-forward pass, which significantly reduces compression time from minutes to seconds.
- 64
- 62
- CATS: Co-saliency Activated Tracklet Selection for Video Co-localization60
This work proposes to leverage co-saliency activated tracklets to address the challenge of jointly localizing common objects across videos and shows that the proposed method outperforms state-of-the-art methods.
- 60
- Explicit Correspondence Matching for Generalizable Neural Radiance Fields58
This work presents a new generalizable NeRF method that is able to directly generalize to new unseen scenarios and perform novel view synthesis with as few as two source views, and considers modeling the cross-view interactions via Transformer cross-attention, which greatly improves the feature matching quality.
- VLIPP: Towards Physically Plausible Video Generation with Vision and Language Informed Physical Prior57
A novel two-stage image-to-video generation framework that explicitly incorporates physics with vision and language informed physical prior and can produce physically plausible motion is proposed.
- Multi-modal Unsupervised Feature Learning for RGB-D Scene Labeling57
This paper adapts the unsupervised feature learning technique for RGB-D labeling as a multi-modality learning problem and performs feature learning and feature encoding simultaneously which significantly boosts the performance.
- Alive Caricature from 2D to 3D55
An algorithm for creating expressive 3D caricatures from 2D caricature images with minimum user interaction is presented, to introduce an intrinsic deformation representation that has the capability of extrapolation, enabling us to create a deformation space from standard face datasets, which maintains face constraints and meanwhile is sufficiently large for producing exaggerated face models.
- 55
- 55
- 54
- 54
- How Far can we Compress Instant-NGP-Based NeRF?53
The Context-based NeRF Compression (CNC) framework is introduced, which leverages highly efficient context models to provide a storage-friendly NeRF representation and excavate both level-wise and dimension-wise context dependencies to enable probability prediction for information entropy reduction.
- 51
- ObjectSDF++: Improved Object-Compositional Neural Implicit Surfaces50
This paper proposes a new framework called ObjectSDF++ to overcome the limitations of ObjectSDF and designs a novel regularization term for object distinction, which can effectively mitigate the issue that ObjectSDF may result in unexpected reconstruction in invisible regions due to the lack of constraint to prevent collisions.
- 50
- 49
- Stitchable Neural Networks48
Stitchable Neural Networks (SN-Net) is presented, a novel scalable and efficient framework for model deployment that cheaply produces numerous networks with different complexity and performance trade-offs given a family of pretrained neural networks, which the authors call anchors.
- JRDB-Pose: A Large-Scale Dataset for Multi-Person Pose Estimation and Tracking48
A thorough experimental study of state-of-the-art multi-person pose estimation and tracking methods on JRDB-Pose is conducted, showing that the dataset imposes new challenges for the existing methods.
- 47
- Robust joint source-channel coding for image transmission over wireless channels47
Experimental results show that the proposed scheme can achieve not only high peak signal-to-noise ratio performance, but also excellent perceptual quality, especially when the channel mismatch occurs.
- Auto-Parsing Network for Image Captioning and Visual Question Answering45
A Probabilistic Graphical Model parameterized by the attention operations on each self-attention layer to incorporate sparse assumption is imposed and a PGM probability-based parsing algorithm is developed by which it can discover what the hidden structure of input is during the inference.
- 45
- Multimodal Transformer with Variable-Length Memory for Vision-and-Language Navigation44
This paper introduces Multimodal Transformer with Variable-length Memory (MTVM) for visually-grounded natural language navigation by modelling the temporal context explicitly and enables the agent to keep track of the navigation trajectory by directly storing previous activations in a memory bank.
- 44
- T-Stitch: Accelerating Sampling in Pre-Trained Diffusion Models with Trajectory Stitching43
This paper introduces sampling Trajectory Stitching T-Stitch, a simple yet efficient technique to improve the sampling efficiency with little or no generation degradation and can be used as a drop-in technique to not only accelerate the popular pretrained stable diffusion models but also improve the prompt alignment of stylized SD models from the public model zoo.
- Sem2NeRF: Converting Single-View Semantic Masks to Neural Radiance Fields43
This work introduces a new task, Semantic-to-NeRF translation, that aims to reconstruct a 3D scene modelled by NeRF, conditioned on one single-view semantic mask as input, and proposes the Sem2NeRF framework, which addresses the highly challenging task by encoding the semantic mask into the latent code that controls the3D scene representation of a pre-trained decoder.
- Disentangled Human Body Embedding Based on Deep Hierarchical Neural Network43
An autoencoder-like network architecture is presented to learn disentangled shape and pose embedding specifically for the 3D human body to improve the reconstruction accuracy and construct a large dataset of human body models with consistent connectivity for the learning of the neural network.
- Learning from the Scene and Borrowing from the Rich: Tackling the Long Tail in Scene Graph Generation43
This paper tackles the issue of scene-object interaction aiming at learning specific knowledge from a scene via an additive attention mechanism and long-tail knowledge transfer which tries to transfer the rich knowledge learned from the head into the tail.
- 42
- 42
- Mesh Snapping: Robust Interactive Mesh Cutting Using Fast Geodesic Curvature Flow42
A fast computation scheme called fast geodesic curvature flow (FGCF), which only needs to solve a smaller and easier problem, is devised here.
- 42
- 41
- Improving Deep Lesion Detection Using 3D Contextual and Spatial Attention41
A dual-attention mechanism to the 3D contextual lesion detection framework is introduced, including the cross-slice contextual attention to selectively aggregate the information from different slices through a soft re-sampling process, and intra-slice spatial attention to focus the feature learning in the most prominent regions.
- Understanding and comparing scalable Gaussian process regression for big data41
It is highlighted that the proper hybrid of global and local scalable GPs may be a promising way to improve both the model capability and scalability for big data.
- Facial Action Unit Detection via Adaptive Attention and Relation40
This work proposes an adaptive attention regression network to regress the global attention map of each AU under the constraint of attention predefinition and the guidance of AU detection, which is beneficial for capturing both specified dependencies by landmarks in strongly correlated regions and facial globally distributed dependencies in weakly correlated regions.
- Transformer Scale Gate for Semantic Segmentation40
This work proposes a simple yet effective module, Transformer Scale Gate (TSG), to optimally combine multi-scale features in transformer-based segmentation models, and exploits cues in self and cross attentions in Vision Transformers for the scale selection.
- On the Selection of Transmission Range in Underwater Acoustic Sensor Networks40
This formulation offers a design guideline for energy-efficient packet transmission operation given a certain network connectivity requirement and seeks analytical modeling to investigate the tradeoff between the energy efficiency and the network connectivity through the selection of the transmission range.
- EcoFormer: Energy-Saving Attention with Linear Complexity39
This work proposes a new binarization paradigm customized to high-dimensional softmax attention via kernelized hashing, called EcoFormer, to map the original queries and keys into low-dimensional binary codes in Hamming space and enables most of the expensive multiply-accumulate operations in attention with simple accumulations to save considerable on-chip energy footprint on edge devices.
- 39
- 39
- Generative Region-Language Pretraining for Open-Ended Object Detection38
This paper forms object detection as a generative problem and proposes a simple framework named GenerateU, which can detect dense objects and generate their names in a free-form way and introduces an evaluation method designed to quantitatively measure the performance of generative out-comes.
- 37
- Cosegmentation of multiple image groups37
Experimental results demonstrate that the proposed multi-group cosegmentation framework is able to discover more accurate object priors and outperform state-of-the-art single-group image cose segmentation methods.
- 37
- Efficient quadtree based block-shift filtering for deblocking and deringing37
This paper integrates quadtree (QT) decomposition with the block-shift filtering for deblocking and extends the proposed QT based block-shifting algorithm for deringi ng JPEG2000 coded images.
- 36
- DifFUSER: Diffusion Model for Robust Multi-Sensor Fusion in 3D Object Detection and BEV Segmentation35
DifFUSER, a novel approach that leverages diffusion models for multi-modal fusion in 3D object detection and BEV map segmentation, is introduced, able to refine or even synthesize sensor features in case of sensor malfunction, thereby improving the quality of the fused output.
- 35
- Accurate and Real-Time 3D Pedestrian Detection Using an Efficient Attentive Pillar Network35
This work introduces a stackable Pillar Aware Attention (PAA) module to enhance pillar feature extraction while suppressing noises in point clouds, and presents Mini-BiFPN, a small yet effective feature network that creates bidirectional information flow and multi-level cross-scale feature fusion to better integrate multi-resolution features.
- Large-Scale Heteroscedastic Regression via Gaussian Process35
This work focuses on the heteroscedastic Gaussian process (HGP) regression that integrates the latent function and the noise function in a unified nonparametric Bayesian framework and develops a variational sparse inference algorithm, named VSHGP, to handle large-scale data sets.
- Diversified and Personalized Multi-Rater Medical Image Segmentation33
This work proposes a two-stage framework named D-Persona, which can provide diversified and personalized results at the same time, achieving new SOTA performance for multi-rater medical image segmentation.
- 33
- An efficient network coding based retransmission algorithm for wireless multicast33
This paper proposes a latency and bandwidth efficient coding algorithm based on the principles of network coding for retransmitting lost packets in a singlehop wireless multicast network and demonstrates its effectiveness over previously proposed network coding based retransmission algorithms.
- 33
- 32
- Towards Unbiased Visual Emotion Recognition via Causal Intervention32
This work proposes a novel Interventional Emotion Recognition Network (IERN) to achieve the backdoor adjustment, which is one fundamental deconfounding technique in causal inference and outperforms state-of-the-art approaches for unbiased visual emotion recognition.
- 32
- GeoConv: Geodesic guided convolution for facial action unit recognition31
A novel geodesic guided convolution (GeoConv) is proposed for AU recognition by embedding 3D manifold information into 2D convolutions, which significantly outperforms the state-of-the-art AU recognition methods.
- Visiting the Invisible: Layer-by-Layer Completed Scene Decomposition31
A system to decompose a scene into individual objects, infer their underlying occlusion relationships, and even automatically learn which parts of the objects are occluded that need to be completed, which demonstrates significant improvement over the state-of-the-art.
- Exploiting Web Images for Weakly Supervised Object Detection31
This paper proposes an easy-to-hard knowledge transfer scheme that incorporates easy web images to provide prior knowledge of object appearance as a good starting point and introduces a sophisticated labor-free method to construct a web dataset with good diversity in object appearance.
- An adaptive RBF-HDMR modeling approach under limited computational budget31
Numerical results reveal that the proposed approach provides more accurate predictions than the original RBF-HDMR with the same computational budget, and the version that uses the maximin sampling criterion and the best-model strategy is a recommended choice.
- 31
- Dynamic Focus-aware Positional Queries for Semantic Segmentation30
This paper proposes a simple yet effective query design for semantic segmentation termed Dynamic Focus-aware Positional Queries (DFPQ), which dynamically generates positional queries conditioned on the cross-attention scores from the preceding decoder block and the positional encodings for the corresponding image features, simultaneously.
- Conditional adversarial synthesis of 3D facial action units30
This paper proposes an AU synthesis framework that combines the well-known 3D Morphable Model (3DMM), which intrinsically disentangles expression parameters from other face attributes, with models that adversarially generate 3DMM expression parameters conditioned on given target AU labels, in contrast to the more conventional approach of generating facial images directly.
- Surveillance video coding via low-rank and sparse decomposition30
Experimental results show that, for the compression of surveillance videos, the proposed scheme can significantly outperform H.264/AVC, up to 3 dB PSNR gain, especially at relatively low bit rates.
- PanSplat: 4K Panorama Synthesis with Feed-Forward Gaussian Splatting29
This paper presents PanSplat, a generalizable, feed-forward approach that efficiently supports resolution up to 4K (2048 × 4096) and proposes a pipeline that integrates a hierarchical spherical cost volume and Gaussian heads with local operations, enabling two-step deferred backpropagation for memory-efficient training on a single A100 GPU.
- Scalable Visual Transformers with Hierarchical Pooling29
A Hierarchical Visual Transformer (HVT) is proposed which progressively pools visual tokens to shrink the sequence length and hence reduces the computational cost, analogous to the feature maps downsampling in Convolutional Neural Networks (CNNs).
- Hierarchical Scene Graph Encoder-Decoder for Image Paragraph Captioning29
This work uses the image scene graph as the "script" to incorporate rich semantic knowledge and, more importantly, the hierarchical constraints into the model by proposing Hierarchical Scene Graph Encoder-Decoder (HSGED) for generating coherent and distinctive paragraphs.
- Watch It Twice29
A novel recurrent video encoding method and a novel visual spatial feature for the video captioning task that encodes the video twice with a predicted key frame to avoid irrelevant temporal information often occurring at the beginning and at the end of a video.
- 29
- Seam carving extension29
This paper applies the recently developed image resizing algorithm, seam carving, for content-aware multi-size image compression and proposes a proposed codec that encodes an image into acontent-aware progressive bitstream that allows decoding into arbitrary display resolution.
- 29
- Vector Quantized Wasserstein Auto-Encoder28
This work endow discrete distributions over sequences of codewords and learn a deterministic decoder that transports the distribution over the sequences ofcodewords to the data distribution via minimizing a WS distance between them to have a better and more controllable clustering solution.
- 28
- 27
- 27
- PCGS: Progressive Compression of 3D Gaussian Splatting26
This work proposes PCGS (Progressive Compression of 3D Gaussian Splatting), which adaptively controls both the quantity and quality of Gaussians (or anchors) to enable effective progressivity for on-demand applications.
- Reliability-Adaptive Consistency Regularization for Weakly-Supervised Point Cloud Segmentation26
A novel Reliability-Adaptive Consistency Network (RAC-Net) is proposed to use both prediction confidence and model uncertainty to measure the reliability of pseudo labels and apply consistency training on all unlabeled points while with different consistency constraints for different points based on the reliability of corresponding pseudo labels.
- Zero-Annotation Object Detection with Web Knowledge Transfer26
This work proposes an object detection method that does not require any form of human annotation on target tasks, by exploiting freely available web images, and introduces a multi-instance multi-label domain adaption learning framework with two key innovations.
- 26
- 25
- Open-Vocabulary Object Detection via Scene Graph Discovery25
This paper proposes a novel Scene-Graph-Based Discovery Network (SGDN) that exploits scene graph cues for OV detection and proposes scene-graph-based prediction (SGPred), where it takes scene graphs as bridges to improve the consistency between cross-modal embeddings for Ov object classification.
- Keypoint based weakly supervised human parsing25
This work develops an iterative learning method to generate pseudo part segmentation masks from keypoint labels and develops a correlation network to perform joint prediction of part and object segmentation masks and improve the segmentation performance.
- Semi-supervised manifold-embedded hashing with joint feature representation and classifier learning25
A semi-supervised manifold embedding is explored to simultaneously optimize feature representation and classifier learning to make the learned binary codes optimal for classification.
- 24
- 23
- 23
- 23
- 22
- 22
- 21
- Pluralistic Free-Form Image Completion21
This paper proposes a novel and probabilistically principled framework with two parallel paths for pluralistic image completion, a reconstructive path that utilizes the only one ground truth to get prior distribution of missing patches and rebuild the original image from this distribution.
- Student-t Process Regression with Student-t Likelihood21
Student-t Process Regression with Student-t Likelihood is theoretically show that TPRT is more robust to both input and target outliers than GPR and GPRT, and it is proved that both G PR and G PRT are special cases of TPRt.
- 21
- Kinect Depth Recovery Using a Color-Guided, Region-Adaptive, and Depth-Selective Framework21
This article formulate their effective features including adaptive support region selection, reliable depth selection, and color guidance together under an optimization framework for Kinect depth recovery as an energy minimization problem, which solves the depth hole filling and denoising simultaneously.
- 21
- 21
- 21
- 21
- 20
- Video captioning with boundary-aware hierarchical language decoding and joint video prediction20
A boundary-aware hierarchical language decoder for video captioning is proposed, which consists of a high-level GRU based languageDecoder, working as a global (caption-level) language model, and a low-levelGRU basedlanguage decoder, worked as a local (phrase-level] language model.
- Wordfence: Text detection in natural images with border awareness20
The concept of WordFence, which detects border areas surrounding each individual word and a novel pixelwise weighted softmax loss function which penalizes background and emphasizes small text regions are presented, producing an end-to-end word detection system.
- 20
- 19
- Primer and Recent Developments on Fountain Codes19
This paper surveys the various erasure codes which had been proposed and patented recently, and along the survey a introductory tutorial on many of the essential concepts and readings in erasure and Fountain codes is provided.
- 19
- McCaD: Multi-Contrast MRI Conditioned, Adaptive Adversarial Diffusion Model for High-Fidelity MRI Synthesis18
McCaD (Multi-Contrast MRI Conditioned Adaptive Adversarial Diffusion), a novel framework leveraging an adversarial diffusion model conditioned on multiple contrasts for high-fidelity MRI synthesis significantly enhances synthesis accuracy by employing a multi-scale, feature-guided mechanism, incorporating denoising and semantic encoders.
- 18
- 18
- Self-supervised relationship probing18
A self-supervised method that implicitly learns the visual relationships without relying on any ground-truth visual relationship annotations is introduced, and the effectiveness of the proposed method on various vision-language tasks that benefit from improved visual relationship understanding is verified.
- 18
- Incremental low-rank and sparse decomposition for compressing videos captured by fixed cameras18
Experimental results show that the proposed coding scheme can significantly improve the existing standard codecs, H.264/AVC and HEVC, and outperform the state-of-the-art background modeling based coding schemes.
- 18
- Online XOR packet coding: Efficient single-hop wireless multicasting with low decoding delay18
This paper presents a cross-layer solution to the problem of unreliability in IEEE 802.11 wireless multicast network, where an Access Point is multicasting a data file to a group of receivers over independent wireless erasure channels and designs an online linear XOR coding algorithm to retransmit the lost packets.
- 17
- 17
- 17
- 17
- 17
- How Well Can Vision Language Models See Image Details?16
This research reveals that incorporating pixel value prediction as one of the VLM pre-training tasks and vision encoder adaptation markedly boosts VLM performance on downstream image-language understanding tasks requiring detailed image perception and video game decision making.
- 16
- 16
- Image Co-Skeletonization via Co-Segmentation16
This article explores a new joint processing topic: image co-skeletonization, which is defined as joint skeleton extraction of the foreground objects in an image collection, and proposes a coupled framework for co-Skeletonization and co-segmentation tasks to facilitate shape information discovery for this process through the co- Segmentation process.
- 16
- 16
- 16
- 16
- 15
- Quadtree Convolutional Neural Networks15
The method decomposes and represents the image as a linear quadtree that is only refined in the non-empty portions of the image, enabling QCNN to learn from sparse images much faster and process high resolution images without the memory constraints faced by traditional CNNs.
- Domain Adaptive Fisher Vector for Visual Recognition15
A Domain Adaptive Fisher Vector (DAFV) method is proposed, which learns a transformation matrix to select the domain invariant components of Fisher vectors and simultaneously solves a regression problem for visual recognition tasks based on the transformed features.
- On Multiple Image Group Cosegmentation15
Experimental results show that the proposed multi-group cosegmentation framework is able to discover more accurate object priors and significantly outperform state-of-the-art single-group image cose segmentation methods.
- 14
- Single-Path Bit Sharing for Automatic Loss-Aware Model Compression14
A simple yet effective method named Single-path Bit Sharing (SBS) for automatic loss-aware model compression, which considers the network pruning as a special case of quantization and provides a unified view for model pruning and quantization.
- Unpaired referring expression grounding via bidirectional cross-modal matching14
A novel bidirectional cross-modal matching (BiCM) framework that introduces top-down perspective via generating query-specific visual attention maps and a knowledge adaptation matching (KAM) module that leverages unpaired training data to adapt pretrained knowledge to the target dataset and task.
- Exploring Bottom-Up and Top-Down Cues With Attentive Learning for Webly Supervised Object Detection14
This work proposes a novel webly supervised object detection (WebSOD) method for novel classes which only requires the web images without further annotations, and combines bottom-up and top-down cues for novel class detection.
- 3DFaceNet: Real-time Dense Face Reconstruction via Synthesizing Photo-realistic Face Images14
A novel face data generation method that renders a large number of photo-realistic face images with different attributes based on inverse rendering and proposes a coarse-to-fine learning framework consisting of three convolutional networks.
- 14
- 14
- 14
- AFFnet - a deep convolutional neural network for the detection of atypical femur fractures from anteriorposterior radiographs13
Accurate AI-based AFF diagnostic software has the potential to improve AFF diagnosis, reduce radiologist error, and allow urgent intervention, thus improving patient outcomes.
- Cross-Adversarial Local Distribution Regularization for Semi-supervised Medical Image Segmentation13
A novel cross-adversarial local distribution (Cross-ALD) regularization is introduced to further enhance the smoothness assumption for semi-supervised medical image segmentation task.
- High-Quality Pluralistic Image Completion via Code Shared VQGAN13
The richness of the representation helps the subsequent deploy- ment of a transformer to effectively learn how to composite and complete a masked image at the discrete code domain, resulting in much better image reconstruction quality.
- Region Deformer Networks for Unsupervised Depth Estimation from Unconstrained Monocular Videos13
A new learning based method consisting of DepthNet, PoseNet and Region Deformer Networks (RDN) to estimate depth from unconstrained monocular videos without ground truth supervision is proposed and can not only achieve the state-of-the-art results on standard benchmarks KITTI and Cityscapes, but also show promising results on a crowded pedestrian tracking dataset.
- Can Partial Strong Labels Boost Multi-label Object Recognition?13
A multi-view multi-instance framework that transforms the multi-label classification problem into a multi-class multi- instance learning problem by extracting object proposals from images and holds the generalization ability to boost the performance of unseen categories by available strong labels.
- 13
- 13
- Advances in Multimedia Information Processing – PCM 201213
This paper introduces an incremental learning approach to video-based face recognition, which efficiently exploits the spatiotemporal information in videos and achieves a 100% recognition rate performed on the Honda/UCSD database, and gives near realtime feedback.
- A novel frame-level bit allocation based on two-pass video encoding for low bit rate video streaming applications13
A frame-level bit allocation scheme based on a latest developed rate-distortion (R-D) model, the �-domain R-D model, for low bit rate streaming applications is proposed.
- 13
- GeReA: Question-Aware Prompt Captions for Knowledge-based Visual Question Answering12
GeReA is proposed, a generate-reason framework that prompts a MLLM like InstructBLIP with question relevant vision and language information to generate knowledge-relevant descriptions and reasons those descriptions for knowledge-based VQA, surpassing all previous state-of-the-art methods on OK-VQA and A-OKVQA datasets.
- End-to-End One-Shot Human Parsing12
An end-to-end human parsing framework that can quickly adapt to the novel classes and mitigate the testing bias issue is proposed, and a contrastive loss at the prototype level is added to enforce inter-class distances.
- Structure-Aware Multimodal Feature Fusion for RGB-D Scene Classification and Beyond12
This article investigates a framework allowing greater spatial flexibility, in which the Fisher vector (FV)-encoded distribution of local CNN features, obtained from a multitude of region proposals per image, is considered instead.
- 12
- Diagnosing state-of-the-art object proposal methods12
This paper conducts the first meta-analysis of various object-level characteristics' impact on state-of-the-art object proposal methods and reveals the limitations of existing methods in terms of non-iconic view, small object size, low color contrast, shape regularity etc.
- Compressive environment matting12
In this paper, a novel environment matting algorithm is proposed to capture and extract the environment matte data effectively and efficiently, and significantly simplifies and accelerates the environmentMatting extraction process while still achieving high-accurate composition results.
- 12
- 12
- 12
- Collision Codes: Decoding Superimposed BPSK Modulated Wireless Transmissions12
A coding scheme that achieves the identification of individual transmission from the synchronized superimposed transmission is devised and a mathematical proof for the existence of such a coding scheme is given.
- 12
- PaRa: Personalizing Text-to-Image Diffusion via Parameter Rank Reduction11
PaRa is proposed, an effective and efficient Parameter Rank Reduction approach for T2I model personalization by explicitly controlling the rank of the diffusion model parameters to restrict its initial diverse generation space into a small and well-balanced target space.
- FocusFormer: Focusing on What We Need via Architecture Sampler11
This paper proposes to learn an architecture sampler to assign higher sampling probabilities to those architectures on the Pareto frontier under different resource constraints during supernet training, making them sufficiently optimized and hence improving their performance.
- 11
- 11
- 11
- 11
- Sharpness-Aware Data Generation for Zero-shot Quantization10
A novel methodology is introduced that takes into account quantized model sharpness in synthetic data generation to enhance generalization and demonstrates the superiority of the proposed method over the state-of-the-art techniques in low-bit quantization settings.
- 10
- Learning to Collocate Visual-Linguistic Neural Modules for Image Captioning10
This work proposes a novel image captioner: learning to Collocate Visual-Linguistic Neural Modules (CVLNM), and makes the following technical contributions: distinguishable module design, a self-attention based module controller for robustifying the visual reasoning, and a part-of-speech based syntax loss imposed on the module Controller for further regularizing the training of the CVLNM.
- Graph-based Group Modelling for Backchannel Detection10
This study uses a graph neural network to model group interaction through all members' implicit and explicit behaviours and achieves the best and second best performance on agreement estimation and backchannel detection tasks, respectively, of the 2022 MultiMediate: Multi-modal Group Behaviour Analysis for Artificial Mediation challenge.
- 10
- Self-supervised CNN for Unconstrained 3D Facial Performance Capture from an RGB-D Camera.10
A novel method for real-time 3D facial performance capture with consumer-level RGB-D sensors that is robust to severe occlusion, fast motion, large rotation, exaggerated facial expressions and diverse lighting and augmenting the training data set in new ways is presented.
- 10
- Kinect-Based Easy 3D Object Reconstruction10
The basic idea is to make use of the existing powerful 2D segmentation tool to refine the silhouette in each color image and then form visual hull via the refined dense silhouettes to improve the 3D object model.
- 10
- 10
- 10
- Layered unequal loss protection with pre-interleaving for fast progressive image transmission over packet-loss channels10
This research proposes to combine the L-ULP with the pre-interleaving to have an improved L-ULP (IL-ULP) system and shows that the proposed IL-ULP scheme even outperforms the global optimal result obtained by any traditional ULP scheme.
- 10
- MU-Diff: a mutual learning diffusion model for synthetic MRI with Application for brain lesions9
This work proposes a mutual learning-based framework with an adversarial diffusion approach that outperforms state-of-the-art methods, delivering accurate lesion synthesis and superior downstream segmentation performance, highlighting the diagnostic value and accuracy of the proposed framework.
- McGrids: Monte Carlo-Driven Adaptive Grids for Iso-Surface Extraction9
The experiment results show that the McGrids can significantly reduce the number of implicit field queries, resulting in significant memory reduction, while producing high-quality meshes with rich geometric details.
- Drive-1-to-3: Enriching Diffusion Priors for Novel View Synthesis of Real Vehicles9
A set of good practices to finetune large pretrained models for a real-world task -- harvesting vehicle assets for autonomous driving applications are consolidated and effective finetuning is led to in a $68.8\% reduction in FID for novel view synthesis over prior arts.
- Efficient Stitchable Task Adaptation9
This work presents a novel framework, Efficient Stitchable Task Adaptation (ESTA), to efficiently produce a palette of fine-tuned models that adhere to diverse resource constraints and surpasses the direct SN-Net adaptation by remarkable margins with significantly lower training time and fewer trainable parameters.
- Half‐body Portrait Relighting with Overcomplete Lighting Representation9
A neural-based model for relighting a half-body portrait image by simply referring to another portrait image with the desired lighting condition is presented, which not only outperforms existing methods for referral-based portrait relighting, but also has the capability generate sequences of relighted images via lighting rotations.
- Recovering facial reflectance and geometry from multi-view images9
This paper presents a lightweight solution for inferring photorealistic facial reflectance and geometry, based on a novel formulation built upon the 3D morphable model for representing 3D textured faces in conjunction with the Blinn-Phong reflection model.
- 9
- Visibility Constrained Generative Model for Depth-Based 3D Facial Pose Tracking9
A statistical 3D morphable model that flexibly describes the distribution of points on the surface of the face model, with an efficient switchable online adaptation that gradually captures the identity of the tracked subject and rapidly constructs a suitable face model when the subject changes is introduced.
- 9
- 9
- Interactive object segmentation from multi-view images9
A non-trivial engineering solution to the problem: given a set of calibrated or un-calibrated multi-view images, by interactively cutting 3~4 images, can the foreground object of the rest images be quickly cutout automatically and accurately?
- 9
- 9
- 9
- 8
- 8
- Modeling Caricature Expressions by 3D Blendshape and Dynamic Texture8
A novel method to model caricature expression is presented, which extends traditional 3DMM representation to caricature domain and a caricature can thus be flexibly deformed into arbitrary expressions with good results visually in both shape and color spaces.
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- Dataset, Challenge, and Evaluation for Tumor Segmentation Variability7
By exploring the two tasks with different metrics, the MMIS-2024 challenge aims to establish a global benchmark for multi-rater medical image segmentation, facilitating clinical AI deployments.
- 7
- 7
- Image Captioning In the Transformer Age7
It is demonstrated that IC still has its specific significance in this age by analyzing the connections between IC with some popular self-supervised learning paradigms and releasing the power of the Transformer architecture that a pre-trained large-scale one can be generalized to various tasks including IC.
- Exploring Uncertainty Measures for Image-caption Embedding-and-retrieval Task7
This study shows that the existing method has poor performance in reliability assessment and investigates another aspect of image-caption embedding-and-retrieval tasks, and proposes posterior uncertainty by considering the retrieval task as a classification task, which can accurately assess the reliability of retrieval results.
- Unsupervised Dense Light Field Reconstruction with Occlusion Awareness7
An unsupervised learning method for LF‐oriented view synthesis, which provides a simple solution for generating quality light fields from a sparse set of views using per‐view disparity as a geometry proxy to warp input views to novel views.
- FaceCollage7
FaceCollage is a robust and real-time system for head reconstruction that can be used to create easy-to-deploy telepresence systems, using a pair of consumer-grade RGBD cameras that provide a wide range of views of the reconstructed user.
- Maximum Multipath Routing Throughput in Multirate Wireless Mesh Networks7
This work provides problem formulation that incorporates collision-free schedule, and then based on this formulation the design of an algorithm with search pruning that jointly optimizes paths and transmission schedule is designed.
- 7
- 7
- 7
- Where and What Matters: Sensitivity-Aware Task Vectors for Many-Shot Multimodal In-Context Learning6
A novel Sensitivity-aware Task Vector insertion framework (STV) is proposed, demonstrating its effectiveness and showing consistent improvements over previous task-vector-based methods with strong generalization.
- Relightable and Dynamic Gaussian Avatar Reconstruction from Monocular Video6
A 3DGS-based human avatar modeling framework, termed as Relightable and Dynamic Gaussian Avatar (RnD-Avatar), that presents accurate pose-variant deformation for high-fidelity geometrical details and achieves state-of-the-art performance in novel view synthesis, novel pose rendering, and relighting.
- 6
- 6
- Photo-realistic Face Images Synthesis for Learning-based Fine-scale 3D Face Reconstruction.6
This paper proposes a novel data generation method by rendering a large number of photo-realistic face images with different attributes based on inverse rendering that can reconstruct 3D face shapes with geometry details from only one image, and demonstrates the robustness of the method to pose, expression and lighting.
- 6
- Cooperative Retransmissions through Collisions6
Compared to traditional ARQ retransmissions, a higher retransmission throughput can be achieved by allowing two interfering senders to cooperatively retransmit selected lost packets at the same time by employing opportunistic listening and analog network coding.
- 6
- Marginalized Generalized IoU (MGIoU): A Unified Objective Function for Optimizing Any Convex Parametric Shapes5
Marginalized Generalized IoU is introduced, a novel loss function that overcomes challenges by projecting structured convex shapes onto their unique shape Normals to compute one-dimensional normalized GIoU and extends MGIoU to MGIoU+ that supports optimizing unstructured convex shapes.
- Large-Scale Data-Free Knowledge Distillation for ImageNet via Multi-Resolution Data Generation5
MUlti-reSolution data-freE (MUSE) is introduced, which generates images at lower resolutions while using Class Activation Maps (CAMs) to ensure that the generated images retain critical, class-specific features, leading to significant performance improvements.
- 5
- GraphITTI: Attributed Graph-based Dominance Ranking in Social Interaction Videos5
This paper aims to reformulate the problem of detecting the Most Dominant Person (MDP) as a person ranking problem by utilizing person-specific attributes such as facial gestures, eye gaze, visual attention and speaking patterns, to inject domain knowledge into the GraphITTI framework.
- 5
- Shading‐Based Surface Recovery Using Subdivision‐Based Representation5
A variational surface reconstruction method built upon the subdivision representations for lighting and geometry is developed, which greatly reduces the number of variables required in surface recovery and intrinsically regularizes the illumination vectors and the surface.
- vCache: Supporting Cost-Efficient Adaptive Bitrate Streaming via NFV-Based Virtual Caching5
A NFV-based virtual cache (vCache) for ABR is designed that can dynamically manages the caching states of each video chunk to achieve overall cost-efficiency, and dynamically provision computing resources for transcoding to guarantee transcoding delays are within an acceptable range.
- Multiple consumer-grade depth camera registration using everyday objects5
This paper proposes a unified framework that jointly computes the optimal extrinsic calibration and depth correction parameters, via replacing checkerboards with everyday objects found in the scene, such as regular furniture.
- Using 3D face priors for depth recovery5
This paper proposes a joint optimization framework that consists of two main steps: transforming the face model for better alignment and applying face priors for improved depth recovery, and demonstrates that the proposed method can achieve up to 23.8% improvement in depth recovery with more accurate face registrations.
- Depth Recovery with Face Priors5
This paper proposes a joint optimization framework that consists of two main steps: deforming the face model for better alignment and applying face priors for improved depth recovery, and demonstrates that the proposed method withFace priors significantly outperforms the baseline method that does not use face prior.
- 5
- 5
- 5
- 5
- 5
- Dynamic Programming-Based Reverse Frame Selection for VBR Video Delivery Under Constrained Resources5
- 5
- 5
- 5
- An Empirical Study on How Video-LLMs Answer Video Questions4
This is the first work to systematically uncover how Video-LLMs internally process and understand video content, offering interpretability and efficiency perspectives for future research.
- 4
- 4
- 4
- 4
- 4
- Learning 3D Human Body Embedding4
This work proposes an autoencoder like network architecture to learn disentangled shape and pose embedding specifically for 3D human body and constructs a large dataset of human body models with consistent topology for the learning of neural network.
- 4
- Multi-class Cosegmentation with Pairwise Active Learning4
A novel dimensionality reduction method is proposed to produce an affinity matrix which reflects both the intrinsic structure of data and the constraints, and correlation clustering is applied on the newly generated affinity matrix to acquire refined results.
- Optimizing content retrieval delay for LT-based distributed cloud storage systems4
It is shown that multiple-stage retrieval of fragments is effective to reduce the content-retrieval delay and can be significantly reduced by optimally scheduling packet requests in a multi-stage fashion.
- 4
- Flexible and Accurate Transparent‐Object Matting and Compositing Using Refractive Vector Field4
The concept of refractive vector is introduced and a piecewise vector field fitting algorithm is introduced to simultaneously remove both noises and errors contained in the extracted matte data to provide great flexibility for transparent‐object environment matting and compositing.
- 4
- Multimedia Networking4
This special issue on multimedia networking presents a fully distributed protocol, scalable island multicast, that effectively integrates IP multicast and application layer multicast for media streaming, which brings in lower end-toend delay lower link stress and lower resource usage than traditional ALM protocols.
- MAC-Layer QoS Management for Streaming Rate-Adaptive VBR Video over IEEE 802.11e HCCA WLANs4
This paper proposes a cross-layer framework for efficiently delivering multiclass rate-adaptive VBR video over HCCA and demonstrates the effectiveness of each individual module and the advantage of dynamic interactions among different modules.
- Illumination and view dependent progressive transmission of semi-regular meshes with subdivision connectivity4
This paper proposes a novel distortion model, which consists of both illumination distortion and geometry distortion, in order to compute the rate-distortion function for each segment of a mesh model in a fast manner.
- 4
- Wireless FGS video transmission using adaptive mode selection and unequal error protection4
An overall end-to-end rate-distortion (R-D) function is constructed that is able to find the optimal solutions for both mode selection and UEP so that an optimal tradeoff between efficiency and robustness can be achieved.
- Layered unequal loss protection for progressive image transmission over packet loss channels4
A layered ULP (L-ULP) scheme for progressive image transmission over packet loss channels is proposed, able to solve the mentioned problems of existing ULP schemes by smartly choosing the layers.
- 4
- Uniform threshold TCQ with block classification for image transmission over noisy channels4
This paper reports the previous study by incorporating the block classification into the integrated scheme, and shows a decent compression ratio and graceful degradation with respect to increasing channel errors.
- 3
- Learning in Order! A Sequential Strategy to Learn Invariant Features for Multimodal Sentiment Analysis3
A multimodal model that is trained either in a single source domain or multiple source domains using a novel and simple sequential learning strategy to train models on videos and texts for multimodal sentiment analysis.
- Deep Semi-supervised Anomaly Detection Using VQ-VAE3
A Semi-Supervised Vector Quantized Variational Autoencoder (ss-VQ-VAE) based on the VQ-VAE architecture is proposed, which leverages discretized latent space embeddings of VQ-VAE to reduce noise and redundancies for better reconstruction of normal data in comparison with anomalous data.
- 3
- 3
- 3
- Robust Survey Aggregation with Student-t Distribution and Sparse Representation3
A robust survey aggregation method based on Student-t distribution and sparse representation that achieves significant improvement over the state-of-the-art methods on both synthetic and real datasets is proposed.
- 3
- 3
- 3
- 3
- 3
- Distributed routing algorithm for low-latency broadcasting in multi-rate wireless mesh network3
A distributed source routing algorithm that exploits the multi-rate feature of the IEEE 802.11 MAC protocol to achieve low latency network-wide broadcasting in WMNs and shows results that are compared to the existing best known results under various network settings.
- 3
- Rate–distortion analysis of leaky prediction based FGS video for constant quality constrained rate adaptation3
An accurate R-D model is proposed, which considers not only the distortion introduced in the current frame and the propagated distortion from the reference frame due to rate adaptation, but also the correlation between them and the corresponding bit allocation algorithm can achieve much more smooth video quality.
- Joint mode selection and unequal error protection for bitplane coded video transmission over wireless channels3
An overall end-to-end rate-distortion (R-D) function, which considers not only the source distortion and the channel distortion introduced in the current frame but also the propagated channel distortion from the previous frames, to find the optimal solutions for both mode selection and UEP.
- 3
- Joint source-channel coding of GGD sources with allpass filtering source reshaping3
A robust joint source-channel coding (RJSCC) scheme based on a combination of OJSCC with allpass filtering source reshaping is proposed, which is able to incorporate both transition probability and bit error rate of a bursty channel model into an end-to-end rate-distortion function.
- A Novel Rate-Distortion Model for Leaky Prediction Based FGS Video Coding3
A novel rate-distortion (R-D) model for leaky prediction based FGS (L-FGS) video coding that matches the actual distortion very well under different channel conditions is proposed.
- Two-pass video encoding for low-bit-rate streaming applications3
Experimental results show the proposed two-pass video encoding scheme is able to achieve not only the improved PSNR results but also much smoother visual quality.
- 2
- Unveiling Effective In-Context Configurations for Image Captioning: An External & Internal Analysis2
This paper conducts a comprehensive external and internal investigation of multimodal in-context learning on the image captioning task, revealing both how ICEs configuration strategies impact model performance through external experiments and characteristic typical patterns through internal inspection.
- Sample-Efficient Learning from Agent Experience2
Compared with classical reinforcement-learning baselines, in-context learning from trial-and-error experience followed by Experience Distillation matches their performance with at least \(9.6\times\) fewer environment samples.
- 2
- CA-OVS: Cluster and Adapt Mask Proposals for Open-Vocabulary Semantic Segmentation2
A Wasserstein-distance-based clustering method is presented to better match masks and classes and it is proposed to transfer the information of the mask proposals from the segmentation dataset to the classification dataset by minimizing their Wasserstein distance.
- 2
- 2
- 2
- 2
- Learning local feature representation from matching, clustering and spatial transform2
Experimental results show the comparable or even better performance than the hand-crafted feature, which sheds lights on learning local feature representation in an unsupervised or weakly supervised manner.
- 3D Unet-based Kidney and Kidney Tumer Segmentation with Attentive Feature Learning2
An improved deep 3D Unet is introduced by enriching the feature representation in CT images using an attention module and achieves 1.5% improvement in the segmentation accuracy when evaluated on the validation set.
- 2
- Weakly-Supervised Attention and Relation Learning for Facial Action Unit Detection2
This paper proposes a novel end-to-end weakly-supervised attention and relation learning framework for AU detection that outperforms the state-of-the-art methods forAU detection, and also achieves superior performance of AU intensity estimation with a simple extension.
- Auto-Encoding Graphical Inductive Bias for Descriptive Image Captioning2
This work proposes Scene Graph Auto-Encoder (SGAE) that incorporates the language inductive bias into the encoder-decoder image captioning framework for more human-like captions and validates the effectiveness of SGAE on the challenging MS-COCOimage captioning benchmark.
- 2
- 2
- 2
- CoDaaS: Content Delivery as a Service for User Generated Contents2
This paper proposes an innovative idea of Content-Delivery-as-a-Service (CoDaaS), which builds on a hybrid media cloud, and offers elastic private virtual content delivery services with an agreed Quality of Service (QoS) to UGC providers.
- 2
- 2
- Rate-smoothed encoding for real-time video streaming applications2
Experimental results show that the proposed joint quality and rate smoothed encoding scheme can make an optimal balance between the quality fluctuation and the rate fluctuation, and hence improve the overall system performance.
- 2
- 2
- 2
- 2
- 2
- Single-Rollout Hidden-State Dynamics for Training-Free RLVR Data Selection1
SHIFT is proposed, a one-shot, training-free selector based solely on inference-time hidden-state dynamics that consistently outperforms training-free diversity and difficulty/uncertainty baselines, improving both in-domain accuracy and transfer to harder evaluation settings.
- 1
- 1
- 1
- 1
- Class Enhancement Losses with Pseudo Labels for Zero-shot Semantic Segmentation1
Novel class enhancement losses are proposed to bypass the use of the background embbedding during training, and simultaneously exploit the semantic relationship between text embeddings and mask proposals by ranking the similarity scores to capture the relationship between seen and unseen classes.
- 1
- 1
- 1
- 1
- Semi-Supervised Unconstrained Action Unit Detection via Latent Feature Domain1
An end-to-end semi-supervised unconstrained AU detection framework, which transfers accurate AU labels from a constrained source domain to an unconStrained target domain by exploiting accurate labels of AU-related facial landmarks, and introduces a novel landmark adversarial loss which can solve the multi-player minimax game in adversarial learning.
- Weakly-Supervised Unconstrained Action Unit Detection via Feature Disentanglement.1
This paper aims to exploit accurate AU labels from a well-constrained source domain for training an AU detector for the target domain of unconstrained in-the-wild images, and generates a new feature domain to combine source-domain facial inner landmark features with target-domain global pose and texture features.
- 1
- 1
- 1
- 1
- 1
- Joint Network Coding for interfering wireless multicast networks1
This paper proposes a Joint Network Coding based Cooperative Retransmission (JNC-CR) scheme, where it is shown that ANC along with DNC can offer a much higher retransmission gain than that attainable through either ANC, DNC or Automatic Repeat reQuest (ARQ) based retransmissions.
- Dynamic scheduling of a mixture of scalable IPTV and VoD traffic over wireless home networks1
The proposed dynamic scheduling algorithm allows VoD traffic to be transmitted not only at a later time but also in advance so as to provide more flexibility to serve the overall traffic to achieve a better QoS.
- Channel Resource Allocation for VoIP Applications in Collaborative IEEE 802.11/802.16 Networks1
An analytical model is developed for the study to show significant improvements in voice capacity for the optimized resource allocation scheme proposed for VoIP applications.
- Medium Access Cooperations for Improving VoIP Capacity over Hybrid 802.16/802.11 Cognitive Radio Networks1
This paper designs a cooperation mechanism for 802.16 and 802.11 to share the same medium with adaptable resource allocation and proposes a simple approach for the adaptation that optimizes the resource allocation.
- 1
- 1
- 1
- Optimal reverse frame selection for stored video delivery under constrained resources1
An optimal reverse frame selection (RFS) algorithm based on dynamic programming for delivering stored video under both bandwidth and buffer size constraints is presented and extended to solve the problem of video delivery over VBR channels where the channel bandwidth is both limited and time-varying.
- Including and optimizing shape parameters in Doo-Sabin subdivision surfaces for interpolation1
A genetic algorithm is developed to compute the optimal set of shape parameters such that a "fairness" measure of the surface is minimized and the effects of the optimal shape parameters on the final interpolatory surfaces are demonstrated.
- 1
- 1
- Constant quality constrained bit allocation for leaky prediction based FGS video streaming1
Experimental results show that the proposed bit allocation algorithm can achieve much more smooth video quality than the traditional uniform bit allocation under both CBR (constant bit rate) and VBR (variable bit rates) channels.
- Efficient Multiview Video Coding Based on MPEG-41
The proposed encoder is very promising for the applications of video-conferencing and 3D telepresence and the improvements come from a more efficient reference structure and the joint estimation of disparity and motion fields in the proposed multiview coding system.
- –
- –
- –
- –
- –
- –
- FastCAT: A framework for fast routing table calculation incorporating multiple protocols–
Experimental evaluations have shown that FastCAT generates routing tables more quickly and accurately than the state-of-the-art routing simulation tool, in a general network of around 200 routers.
- Dynamic data replication for distributed cloud storage–
This thesis investigates a cost optimization problem for data replication in distributed cloud storage with two different settings: offline and online and proposes two distributed algorithms: a 3-competitive basic algorithm and a dynamic algorithm that are lightweight and easy to implement.
- –
- "Deepfake" Portrait Image Generation–
This talk mainly review general image generation techniques, particularly describing a few of the authors' recent work on high-quality portrait/facial image generation and presenting a framework for pluralistic facial image generation from a masked facial input.
- MED-TEX: Transferring and Explaining Knowledge with Less Data from Pretrained Medical Imaging Models–
- –
- –
- –
- Search Guided Saliency–
A new type of saliency as inspired by findings from visual search studies - the searching difficulty is correlated with the target-distractor contrast, the distractor homogeneity, as well as the target uniqueness.
- Message from ICME'12 Technical Program Chairs–
The ICME 2012 daily technical program has been designed in such a way that ICME audience will not miss the keynotes, overview talks and the plenary session, and can always find a high quality talk of interest at the parallel oral sessions.
- –
- Nonlinear image restoration using recurrent radial basis function network–
Simulation results show that the proposed recurrent RBFN scheme outperforms the existing RBFN based methods in both visual quality and complexity when the degraded process is recursive.
- –
- IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY Editor-in-Chief–
This special issue, which focuses on event analysis in broad problem domains, has witnessed the effectiveness of using both static and temporal information in event recognition from other video sources.
- –
- –
- Improving scalable video transmission over IEEE 802.11e networks–
This paper designs an appropriate cross-layer video adaptation mechanism for the scalable video that further improves the video quality combining with the proposed resource allocation technique and shows improvement in video quality and resource usage when the proposed technique is implemented.
- The Semantics of Structured Specifications Based on Meta-Design Pattern–
The semantics to meet the structured requirements for vari- ous meta-design pattern with the constructive specification method is discussed and an approach to describe the design patterns based on meta- design pattern is provided.
- Ubiquitous map-image access through wireless overlay networks–
A vertical handoff management system based on mobile IPv6, which can automatically manage the multiple network interfaces on the mobile device, and make decisions on network interface selection according to the current situation is proposed.
- Layered unequal loss protection for image transmission over packet loss channels with delay constraints–
A layered ULP (L-ULP) scheme for fast and efficient FEC allocations among different packet groups and also within each packet group is proposed.
- Proxy-based handheld device access to live NASA satellite weather images–
A new proxy-based system to allow public use handheld devices to access instantly to the Low Resolution Picture Taking satellite weather image data and display regional weather image is presented.
- –
- FEC-Based Wireless Video Streaming with Pre-Interleaving–
This research proposes a novel robust video streaming system in which an interleaving is applied to the compressed bitstream before the channel coding, able to facilitate both an efficient channel coding and an efficient source coding for robust wireless video streaming.
- Joint source-channel coding with allpass filtering source shaping for image transmission over noisy channels–
Compared with the state-of-the-art JSCC schemes, this proposed scheme outperforms most of them especially when the channel mismatch occurs, and is integrated with allpass filtering source shaping to further improve the robustness against channel errors.
- Uniform-threshold TCQ with block classification for image transmission over noisy channels–
This paper reports the recent study by incorporating the block classification into the integrated scheme, which shows decent compression ratio and gracious degradation with respect to increasing channel errors.
- ON ASYNCHRONOUS SEQUENTIAL CIRCUITS WITH FAULT SELF-DETECTION CAPABILITIES–
A state assignment, by which a fault of the circuits may be selfdetected, is presented in case the fault-pattern is known, the fault can be easily detected in a real-time processing environment.
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 2026-10-12. 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.