Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks134 from public data
- TransReID: Transformer-based Object Re-Identification1,342
This work first encode an image as a sequence of patches and build a transformer-based strong baseline with a few critical improvements, which achieves competitive results on several ReID benchmarks with CNN-based methods.
- Hallucination of Multimodal Large Language Models: A Survey466
A comprehensive analysis of the phenomenon of hallucination in multimodal large language models (MLLMs), also known as Large Vision-Language Models (LVLMs), offering a detailed overview of the underlying causes, evaluation benchmarks, metrics, and strategies developed to address this issue.
- MHFormer: Multi-Hypothesis Transformer for 3D Human Pose Estimation452
A Multi-Hypothesis Transformer (MHFormer) that learns spatio-temporal representations of multiple plausible pose hypotheses and achieves state-of-the-art results on two challenging datasets: Human3.6M and MPI-INF-3DHP.
- RGB-D-based human motion recognition with deep learning: A survey403
A detailed overview of recent advances in RGB-D-based motion recognition is presented, and the methods of encoding spatial-temporal-structural information inherent in video sequence are highlighted, and potential directions for future research are discussed.
- Action Recognition Based on Joint Trajectory Maps Using Convolutional Neural Networks403
A compact, effective yet simple method to encode spatio-temporal information carried in 3D skeleton sequences into multiple 2D images, referred to as Joint Trajectory Maps (JTM), and ConvNets are adopted to exploit the discriminative features for real-time human action recognition.
- 351
- 315
- Exploiting Temporal Contexts With Strided Transformer for 3D Human Pose Estimation314
An improved Transformer-based architecture, called Strided Transformer, which simply and effectively lifts a long sequence of 2D joint locations to a single 3D pose, and achieves state-of-the-art results with fewer parameters.
- RGB-D-based action recognition datasets: A survey291
A comprehensive review of the most commonly used action recognition related RGB-D video datasets, including 27 single-view datasets, 10 multi- view datasets, and 7 multi-person datasets, resulting in a number of recommendations for collection of new datasets and use of evaluation protocols.
- 289
- PoseFormerV2: Exploring Frequency Domain for Efficient and Robust 3D Human Pose Estimation257
PoseFormerV2 is proposed, which exploits a compact representation of lengthy skeleton sequences in the frequency domain to efficiently scale up the receptive field and boost robustness to noisy 2D joint detection.
- Action recognition based on joint trajectory maps with convolutional neural networks237
An effective yet simple method to represent spatio-temporal information carried in 3 D skeleton sequences into three 2 D images by encoding the joint trajectories and their dynamics into color distribution in the images, referred to as Joint Trajectory Maps (JTM).
- EPro-PnP: Generalized End-to-End Probabilistic Perspective-n-Points for Monocular Object Pose Estimation229
EPro-PnP is proposed, a probabilistic PnP layer for general end-to-end pose estimation, which outputs a distribution of pose with differentiable probability density on the SE(3) manifold, and helps to explore new possibilities of network design.
- Selective Structured State-Spaces for Long-Form Video Understanding200
A novel Selective S4 model that employs a lightweight mask generator to adaptively select informative image tokens resulting in more efficient and accurate modeling of long-term spatiotemporal dependencies in videos and a novel long-short masked contrastive learning (LSMCL) approach that enables the model to predict longer temporal context using shorter input videos.
- Skeleton-based action recognition using LSTM and CNN192
The proposed method achieved state-of-the-art results on NTU RGB+D datasets for 3D human action analysis and ranked 1st place in Large Scale 3D Human Activity Analysis Challenge in Depth Videos.
- Zen-NAS: A Zero-Shot NAS for High-Performance Image Recognition186
A novel zero-shot index dubbed Zen- score is proposed to rank the architectures in Neural Architecture Search by maximizing the Zen-Score of the target network under given inference budgets, which is magnitude times faster on multiple server-side and mobile-side GPU platforms with state-of-the-art accuracy on ImageNet.
- Depth Pooling Based Large-Scale 3-D Action Recognition With Convolutional Neural Networks184
The image-based representations enable the existing ConvNet models trained on image data without training a large number of parameters from scratch, and achieve the state-of-art results on three large datasets, namely, the large-scale continuous gesture recognition dataset.
- KVT: k-NN Attention for Boosting Vision Transformers159
A sparse attention scheme, dubbed k-NN attention, which naturally inherits the local bias of CNNs without introducing convolutional operations, and allows for the exploration of long range correlation and filter out irrelevant tokens by choosing the most similar tokens from the entire image.
- Scene Flow to Action Map: A New Representation for RGB-D Based Action Recognition with Convolutional Neural Networks157
A new representation, namely, Scene Flow to Action Map (SFAM), that describes several long term spatio-temporal dynamics for action recognition from RGB-D data and takes better advantage of the trained ConvNets models over ImageNet.
- One Token to Seg Them All: Language Instructed Reasoning Segmentation in Videos147
VideoLISA is introduced, a video-based multimodal large language model designed to tackle the problem of language-instructed reasoning segmentation in videos, and shows promising generalization to image segmentation, revealing its potential as a unified foundation model for language-instructed object segmentation.
- Unsupervised feature selection via latent representation learning and manifold regularization144
A robust unsupervised feature selection method which embeds the latent representation learning into feature selection and is carried out in the learned latent representation space which is more robust to noises.
- TransRPPG: Remote Photoplethysmography Transformer for 3D Mask Face Presentation Attack Detection142
This letter proposes a pure rPPG transformer (TransRPPG) framework for learning intrinsic liveness representation efficiently and is lightweight and efficient, which is promising for mobile-level applications.
- Robust unsupervised feature selection via dual self-representation and manifold regularization138
Experimental results on ten real-world data sets demonstrate that the proposed method can effectively identify important features, outperforming many state-of-the-art unsupervised feature selection methods in terms of clustering accuracy (ACC) and normalized mutual information (NMI).
- Searching Multi-Rate and Multi-Modal Temporal Enhanced Networks for Gesture Recognition137
The proposed method includes two key components: enhanced temporal representation via the proposed 3D Central Difference Convolution (3D-CDC) family, which is able to capture rich temporal context via aggregating temporal difference information; and optimized backbones for multi-sampling-rate branches and lateral connections among varied modalities.
- ConvNets-Based Action Recognition from Depth Maps through Virtual Cameras and Pseudocoloring112
Through transferring learning the models originally trained over ImageNet for image classification, the three ConvNets are trained independently on the color-coded DMMs constructed in three orthogonal planes, achieving state-of-the-art results on these datasets.
- 111
- 109
- Trear: Transformer-Based RGB-D Egocentric Action Recognition107
A transformer-based RGB-D egocentric action recognition framework that adopts a self-attention mechanism to model the temporal structure of the data from different modalities and outperforms the state-of-the-art results by a large margin.
- Consensus learning guided multi-view unsupervised feature selection107
A consensus learning guided multi-view unsupervised feature selection method, which embeds multi-View feature selection into a non-negative matrix factorization based clustering with sparse constrain.
- Generation of stomach tissue from mouse embryonic stem cells107
In vitro stomach tissue derived from pluripotent stem cells mimics in vivo development and can be used for stomach disease models.
- Large-scale Isolated Gesture Recognition using Convolutional Neural Networks100
A convolutional Neural networks (ConvNets) based method is developed for gesture recognition and evaluated on the Large-scale Isolated Gesture Recognition at the ChaLearn Looking at People (LAP) challenge 2016.
- Investigation of different skeleton features for CNN-based 3D action recognition95
This paper proposes to encode five spatial skeleton features into images with different encoding methods and achieves state-of-the-art performance on NTU RGB+D dataset for 3D human action analysis.
- 94
- 92
- 86
- Text Is MASS: Modeling as Stochastic Embedding for Text-Video Retrieval85
A new stochastic text modeling method T-MASS is proposed, i.e., text is modeled as a stochastic embedding to enrich text embedding with a flexible and re-silient semantic range, yielding a text mass.
- Robust graph regularized unsupervised feature selection85
This work proposes a unified UFS framework via feature self-representation and robust graph regularization, with the aim at reducing the sensitivity to outliers from the following two aspects: an l2, 1-norm is used to characterize the feature representation residual matrix and an l1-norm based graph Laplacian regularization term is adopted to preserve the local geometric structure of data.
- The Amazon Nova Family of Models: Technical Report and Model Card71
A new generation of state-of-the-art foundation models that deliver frontier intelligence and industry-leading price performance are presented, and benchmarking results for core capabilities, agentic performance, long context, functional adaptation, runtime performance, and human evaluation are reported.
- 71
- Hourglass Tokenizer for Efficient Transformer-Based 3D Human Pose Estimation66
This paper proposes a token pruning cluster (TPC) that dynamically selects a few representative tokens with high semantic diversity while eliminating the redundancy of video frames and develops a token recovering attention (TRA) to restore the detailed spatio-temporal information based on the selected tokens, thereby expanding the network output to the original full-length temporal resolution for fast inference.
- 66
- 61
- Scaled ReLU Matters for Training Vision Transformers59
It is verified, both theoretically and empirically, that scaled ReLU in conv-stem not only improves training stabilization, but also increases the diversity of patch tokens, thus boosting peak performance with a large margin via adding few parameters and flops.
- VTC-LFC: Vision Transformer Compression with Low-Frequency Components59
Two metrics named low- frequencies sensitivity (LFS) and low-frequency energy (LFE) and a bottom-up cascade pruning scheme are applied to compress different dimensions jointly and demonstrate that the proposed method could save 40% ∼ 60% of the FLOPs in ViTs, thus increasing the throughput on practical devices with less than 1% performance drop on ImageNet-1K.
- Reconstruction of renal glomerular tissue using collagen vitrigel scaffold59
The coculture results showed that the collagen vitrigel maintains cell growth and cell viability for more than 1 month, and surprisingly, the epithelial layer demonstrated polarity formation, which usually appears in in vivo normal epithelial cells existing at the glomerular basement membrane, but seldom appears in epithel cells cultured in vitro.
- 58
- Making Vision Transformers Efficient from A Token Sparsification View56
A novel Semantic Token ViT (STViT) is proposed, for efficient global and local vision transformers, which can also be revised to serve as backbone for downstream tasks, and a STViT-R(ecovery) network is designed to restore the detailed spatial information based on the ST ViT, making it work for upstream tasks, which is powerless for previous token sparsification methods.
- Large-scale Continuous Gesture Recognition Using Convolutional Neural Networks55
This paper addresses the problem of continuous gesture recognition from sequences of depth maps using Convolutional Neural networks (ConvNets) and first segments individual gestures from a depth sequence based on quantity of movement (QOM).
- Audio-Enhanced Text-to-Video Retrieval using Text-Conditioned Feature Alignment53
TEFAL is introduced, a TExt-conditioned Feature ALignment method that produces both audio and video representations conditioned on the text query that achieves better than state-of-the-art performance consistently across the four datasets.
- 52
- TransFGU: A Top-Down Approach to Fine-Grained Unsupervised Semantic Segmentation51
This work proposes the first top-down unsupervised semantic segmentation framework for fine-grained segmentation in extremely complicated scenarios and shows that it is robust to both object-centric and scene-centric datasets under different semantic granularity levels, and outperforms all the current state-of-the-art bottom-up methods.
- Decoupling and Recoupling Spatiotemporal Representation for RGB-D-based Motion Recognition51
Seamless combination of these novel designs forms a robust spatiotemporal representation and achieves better performance than state-of-the-art methods on four public motion datasets.
- 50
- Online human action recognition based on incremental learning of weighted covariance descriptors49
An effective framework of such an approach to online action recognition from a stream of noisy skeleton data, using a weighted covariance descriptor as a means to accumulate information is presented.
- Augmented Transformer with Adaptive Graph for Temporal Action Proposal Generation47
This work enhances the vanilla transformer by equipping a snippet actionness loss and a front block, dubbed augmented transformer, and it improves the abilities of capturing long-range dependencies and learning robust feature for noisy action instances, and outperforms state-of-the-art TAPG methods.
- 46
- Mining Mid-Level Features for Action Recognition Based on Effective Skeleton Representation45
An effective method is proposed to extract mid-level features from Kinect skeletons for 3D human action recognition and this new representation yields state-of-the-art results on MSR DailyActivity3D and MSR ActionPairs3D.
- 44
- 44
- Saliency detection via affinity graph learning and weighted manifold ranking42
Comprehensive evaluations on three challenge datasets indicate that the proposed bottom-up saliency detection approach by affinity graph learning and weighted manifold ranking universally surpasses other unsupervised graph based saliency Detection methods.
- Development of a human corneal epithelium model utilizing a collagen vitrigel membrane and the changes of its barrier function induced by exposing eye irritant chemicals40
The results suggest that the TEER assay using the human corneal epithelial model is very useful for an ocular irritancy evaluation as an alternative to the Draize eye irritation test.
- Exploiting Better Feature Aggregation for Video Object Detection37
The proposed feature alignment module is a correlation based feature alignment method to align the support and target frames for feature aggregation in the temporal domain that improves the accuracy of single-frame detectors significantly, and outperforms previous temporal or spatial relation networks.
- 37
- A Unified Multimodal De- and Re-Coupling Framework for RGB-D Motion Recognition36
A novel video data augmentation method dubbed ShuffleMix is introduced, which acts as a supplement to MixUp, to provide additional temporal regularization for motion recognition and a novel cross-modal Complement Feature Catcher is explored to improve the late fusion results.
- BP-triplet net for unsupervised domain adaptation: A Bayesian perspective35
This work proposes an effective BP-Triplet Loss for unsupervised domain adaption (UDA) from the perspective of Bayesian learning and it can self attend to the hard pairs and can be upper bounded following Ben-David s theorem.
- 35
- A Hybrid Network for Large-Scale Action Recognition from RGB and Depth Modalities35
The paper presents a novel hybrid network that effectively leverages the strengths of the emerging Convolutional Neural Network and Recurrent Neural Network based approaches to specifically address the challenges that occur in large-scale action recognition and are not fully dealt with by the state-of-the-art methods.
- 34
- SCT: A Simple Baseline for Parameter-Efficient Fine-Tuning via Salient Channels32
Experiments outperform full fine-tuning on 18 out of 19 tasks in the VTAB-1K benchmark by adding only 0.11M parameters of the ViT-B, demonstrating the proposed tuning technique’s strong capability and effectiveness in the low-data regime.
- Transformer guided geometry model for flow-based unsupervised visual odometry32
A method consisting of two camera pose estimators that deal with the information from pairwise images and a short sequence of images, respectively that outperforms the state-of-the-art unsupervised learning-based methods by a large margin and performs comparably to supervised and traditional ones on the KITTI and Malaga dataset.
- R²MRF: Defocus Blur Detection via Recurrently Refining Multi-Scale Residual Features29
A deep neural network which Recurrently Refines Multi-scale Residual Features (R2MRF) for defocus blur detection consistently and significantly outperforms the competitors in terms of both efficiency and accuracy.
- A Review of Dynamic Maps for 3D Human Motion Recognition Using ConvNets and Its Improvement27
A novel encoding scheme is developed that generates effective flow-guided dynamic maps, and they could select the high motion window and distinguish the order among the frames with small motion on the large Chalearn LAP IsoGD and NTU RGB+D datasets.
- An effective edge-preserving smoothing method for image manipulation26
The method provides an unified framework to regularize the edge and texture pixels in the optimization so that geometric edges representing image structures can be well retained and fine edges of texture regions are removed or suppressed.
- A Large Scale RGB-D Dataset for Action Recognition24
Four state-of-the-art algorithms are evaluated on the large combined dataset and the results have verified the limitations of current algorithms and the effectiveness of the large dataset.
- Human Pose-based Estimation, Tracking and Action Recognition with Deep Learning: A Survey23
This survey presents a comprehensive survey of pose-based applications utilizing deep learning, encompassing pose estimation, pose tracking, and action recognition, and emphasizes methodologies for integrating these three tasks into a unified framework within video sequences.
- Training-Free Text-Guided Image Editing with Visual Autoregressive Model21
This work proposes a noel text-guided image editing framework based on VAR (Visual AutoRegressive modeling), which eliminates the need for explicit inversion while ensuring precise and controlled modifications, and introduces a caching mechanism that stores token indices and probability distributions from the original image.
- Multi-stage Factorized Spatio-Temporal Representation for RGB-D Action and Gesture Recognition21
The seamless integration of these innovative designs results in a robust spatio-temporal representation that outperforms state-of-the-art approaches on RGB-D action and gesture recognition datasets.
- Spatially and Temporally Structured Global to Local Aggregation of Dynamic Depth Information for Action Recognition21
This new representation, referred to as Spatially and Temporally Structured Dynamic Depth Images, aggregates from global to fine-grained levels motion and structure information in a depth sequence, and enables the existing ConvNet models trained on image data for classification of depth sequences, without a need for training the models afresh.
- Context and Structure Mining Network for Video Object Detection19
The Context and Structure Mining Network is proposed to better aggregate features for video object detection and outperforms all the latest state-of-the-art methods on the ImageNet VID dataset with a large margin.
- SAR-NAS: Skeleton-based action recognition via neural architecture searching19
This paper encode a skeleton-based action instance into a tensor and carefully define a set of operations to build two types of network cells: normal cells and reduction cells, to search for an effective network architecture that is built upon the two kinds of cells.
- Frequency Domain Disentanglement for Arbitrary Neural Style Transfer17
A novel Frequency Domain Disentanglement (FDD) framework for arbitrary neural style transfer based on the FreMixer module that disentangles and re-entangles the frequency spectrum of content and style components in the frequency domain.
- 17
- 17
- 15
- 15
- A novel rate control algorithm for video coding based on fuzzy-PID controller14
A self-parameter-tuning fuzzy-PID controller is proposed to reduce the deviation between the target buffer level and the current buffer fullness and achieves smooth target bits while enabling better buffer control and visual quality.
- Synthesis of porphyrin-incorporated polymers and their application for simultaneous detection of multimetal components by using spectrophotometry13
Altering of the spectrophotometrical metal sensitivity of porphyrin was attempted and this method was applied to detect multimetal components in water samples such as tap water and industrial effluent.
- Beyond Speaker Identity: Text Guided Target Speech Extraction12
Experimental results demonstrate that the proposed text-guided TSE model StyleTSE effectively separates speech based not only on who is speaking, but also on how they are speaking, enhancing TSE when traditional audio clues are absent.
- Salient object detection using color spatial distribution and minimum spanning tree weight12
This paper uses a segmentation algorithm to decompose an image into superpixel-level elements, then uses these elements as nodes to construct a minimum spanning tree (MST), each connected edge weight is the mean color difference between two nodes.
- An Improved Direction Finding Algorithm Based on Toeplitz Approximation12
In this paper, a novel direction of arrival (DOA) estimation algorithm called the Toeplitz fourth order cumulants multiple signal classification method (TFOC-MUSIC) algorithm is proposed through combining a fast MUSIC-like algorithm termed the modified fourth order Cumulants MUSIC (MFOC- MUSic) algorithm and ToePlitz approximation and to improve the estimation performance.
- An Improved Direction Finding Algorithm Based on12
In this paper, a novel direction of arrival (DOA) estimation algorithm called the Toeplitz fourth order cumulants multiple signal classification method (TFOC-MUSIC) algorithm is proposed through combining a fast MUSIC-like algorithm termed the modified fourth order Cumulants MUSIC (MFOC- MUSic) algorithm and ToePlitz approximation and to improve the estimation performance.
- Factorized Visual Tokenization and Generation11
Factorized Quantization (FQ) is introduced, a novel approach that revitalizes VQ-based tokenizers by decomposing a large codebook into multiple independent sub-codebooks, enabling more efficient and scalable visual tokenization.
- 11
- H 2 OT: Hierarchical Hourglass Tokenizer for Efficient Video Pose Transformers10
H2OT reveals that maintaining the full pose sequence is unnecessary, and a few pose tokens of representative frames can achieve both high efficiency and estimation accuracy, and is easily incorporated into common VPT models on both seq2seq and seq2frame pipelines.
- 9
- Detection of initial angiogenesis from dorsal aorta into metanephroi and elucidation of its role in kidney development9
It is suggested that blood flow and dynamic circulation of various factors at the early developing stage may be prerequisite to a successful construction of blood vessels in the complicated organs either in vitro or in vivo.
- Improved MTT(3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) assay for the measurement of viable animal cell number in porous cellulose carriers9
Results of the comparison of 5 colorimetric assays showed that MTT(3-(4,5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromide) assay was the most suitable method for the measurement of cell concentration in porous cellulose carriers.
- 8
- Revisiting Vision Transformer from the View of Path Ensemble8
This work equivalently transforms the traditional cascade of multi-head self-attention and feed-forward network into three parallel paths in each transformer layer, and further transforms the ViT into an explicit multi-path ensemble network.
- Effective Vision Transformer Training: A Data-Centric Perspective8
This paper proposes a novel data-centric ViT training framework to dynamically measure the ``difficulty'' of training samples and generate ``effective'' samples for models at different training stages, and proposes a patch-level erasing strategy dubbed PatchErasing.
- 8
- Revisit Parameter-Efficient Transfer Learning: A Two-Stage Paradigm7
A novel two-stage paradigm, where the pre-trained model is first aligned to the target distribution and the task-relevant information is leveraged for effective adaptation, achieves state-of-the-art performance on the average accuracy of 19 downstream tasks.
- FT-HID: a large-scale RGB-D dataset for first- and third-person human interaction analysis7
A novel multi-view interaction mechanism for skeleton sequences, and a joint learning multi-stream framework for first-person and third-person visions are introduced, which yield promising results on the FT-HID dataset.
- DOAD: Decoupled One Stage Action Detection Network6
This work presents a decoupled onestage network dubbed DOAD, to mitigate above issues and improve the efficiency for spatio-temporal action detection and shows that the method is competitive with two-stage state-of-theart methods with significant efficiency improvement.
- 5
- Adaptive Query Selection for Camouflaged Instance Segmentation4
This work designs a boundary location embedding mechanism that incorporates fake instance boundaries to obtain better location information for more accurate query instance matching and significantly improves the performance on two challenging camouflaged instance segmentation datasets.
- 4
- Ctrl&Shift: High-Quality Geometry-Aware Object Manipulation in Visual Generation3
This work presents Ctrl&Shift, an end-to-end diffusion framework to achieve geometry-consistent object manipulation without explicit 3D representations, the first framework to unify fine-grained geometric control and real-world generalization for object manipulation, without relying on any explicit 3D modeling.
- 3
- 2
- Align2Concept: Language Guided Interpretable Image Recognition by Visual Prototype and Textual Concept Alignment2
This work proposes a Prototype-Concept Alignment Network (ProCoNet) for learning visual prototypes under the guidance of textual concepts and shows that the ProCoNet provides higher accuracy and better interpretability compared to the single-modality interpretable model.
- Image-to-Video Re-Identification via Mutual Discriminative Knowledge Transfer2
This paper proposes the triplet contrast loss (TCL), a novel loss designed for KD that selectively transfers the local discriminative features from teacher to student, making it effective in the ReID.
- 2
- Learning attentive dynamic maps (ADMs) for Understanding Human Actions2
A novel end-to-end trainable deep architecture to learn an attentive dynamic map (ADM) for understanding human motion from skeleton data achieved competitive performance compared with the state-of-the-art results on the Large Scale Combined dataset, MSRC-12 dataset and NTU RGB+D dataset.
- Physical and biological studies on DNA/anti-DNA immune complex formed in vitro2
In vitro preparation of DNA/anti-DNA immune complex using monoclonal antibody and low molecular weight homogeneous DNA was described and the characteristics of the immune complex were identified and studies of organ distribution and uptake of IC demonstrated a particular affinity to the kidney tissue.
- 1
- 1
- 1
- 1
- 1
- Look-Before-Move: Narrative-Grounded World Visual Attention in Dynamic 3D Story Worlds–
Look-Before-Move, a camera planning framework that separates observation specification from motion execution, is proposed and experiments show that the framework improves subject perception, intent consistency, and trajectory quality over representative baselines, demonstrating the importance of organizing visual attention before generating camera motion.
- –
- Social Structure Matters in 3D Human-Human Interaction Generation–
The Solo-to-Social framework bridges social organization and motion realization, producing 3D HHI with improved phase consistency, role alignment, and partner-aware coordination.
- –
- –
- –
- Unraveling Movie Genres through Cross-Attention Fusion of Bi-Modal Synergy of Poster–
This paper presented the framework that exploits movie posters from a visual and textual perspective to address the multilabel movie genre classification problem and introduced a cross-attention-based fusion module to allocate attention weights to visual and textual embedding.
- –
- Action recognition from RGB-D data–
To exploit the joint information in multi-modal features arising from heterogeneous sources (RGB, depth), it is proposed to cooperatively train a single ConvNet on both RGB features and depth features, and deeply aggregate the two modalities to achieve robust action recognition.
- PID-based bit allocation strategy for asymmetric stereoscopic video coding with fuzzy quality controller–
A bit allocation strategy for asymmetric stereoscopic video coding is proposed and a proportionalintegral-derivative controller is adopted in order to improve the accuracy of bit allocation and rate control in the left view, and fewer bits are allocated for the right view so that the overall bit rate is saved.
- –
- –
- –
- –
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.