Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks299 from public data
- 2,618
- Learning Spatio-Temporal Representation with Pseudo-3D Residual Networks1,874
This paper devise multiple variants of bottleneck building blocks in a residual learning framework by simulating 3 x3 x 3 convolutions with 1 × 3 × 3 convolutional filters on spatial domain (equivalent to 2D CNN) plus 3 × 1 × 1 convolutions to construct temporal connections on adjacent feature maps in time.
- Exploring Visual Relationship for Image Captioning933
This paper introduces a new design to explore the connections between objects for image captioning under the umbrella of attention-based encoder-decoder framework that novelly integrates both semantic and spatial object relationships into image encoder.
- Contextual Transformer Networks for Visual Recognition749
A novel Transformer-style module, i.e., Contextual Transformer (CoT) block, for visual recognition that fully capitalizes on the contextual information among input keys to guide the learning of dynamic attention matrix and thus strengthens the capacity of visual representation.
- Boosting Image Captioning with Attributes733
This paper presents Long Short-Term Memory with Attributes (LSTM-A) - a novel architecture that integrates attributes into the successful Convolutional Neural Networks plus Recurrent Neural Networks (RNNs) image captioning framework, by training them in an end-to-end manner.
- X-Linear Attention Networks for Image Captioning684
A unified attention block --- X-Linear attention block, that fully employs bilinear pooling to selectively capitalize on visual information or perform multi-modal reasoning is introduced.
- Jointly Modeling Embedding and Translation to Bridge Video and Language563
A novel unified framework, named Long Short-Term Memory with visual-semantic Embedding (LSTM-E), which can simultaneously explore the learning of LSTM and visual- semantic embedding and outperforms several state-of-the-art techniques in predicting Subject-Verb-Object (SVO) triplets.
- Transferrable Prototypical Networks for Unsupervised Domain Adaptation398
This paper presents Transferrable Prototypical Networks (TPN), which is end-to-end trained by jointly minimizing the distance across the prototypes on three types of data and KL-divergence of score distributions output by each pair of the prototypes.
- Fully Convolutional Adaptation Networks for Semantic Segmentation375
FCAN is presented, a novel deep architecture for semantic segmentation which combines Appearance Adaptation Networks (AAN) and Representation Adaptation networks (RAN), which learns a transformation from one domain to the other in the pixel space and RAN is optimized in an adversarial learning manner to maximally fool the domain discriminator with the learnt source and target representations.
- Gaussian Temporal Awareness Networks for Action Localization369
Gaussian Temporal Awareness Networks (GTAN) is presented --- a new architecture that novelly integrates the exploitation of temporal structure into an one-stage action localization framework and achieves superior results when comparing to state-of-the-art approaches.
- Video Captioning with Transferred Semantic Attributes364
This paper presents Long Short-Term Memory with Transferred Semantic Attributes (LSTM-TSA), a novel deep architecture that incorporates the transferred semantic attributes learnt from images and videos into the CNN plus RNN framework, by training them in an end-to-end manner.
- Exploring Object Relation in Mean Teacher for Cross-Domain Detection355
This work presents Mean Teacher with Object Relations (MTOR) that novelly remolds Mean Teacher under the backbone of Faster R-CNN by integrating the object relations into the measure of consistency cost between teacher and student modules.
- 330
- Memory Matching Networks for One-Shot Image Recognition310
Memory Matching Networks (MM-Net) is presented - a novel deep architecture that explores the training procedure, following the philosophy that training and test conditions must match, and improves one-shot accuracy on Omniglot and miniImageNet.
- 283
- Relation Distillation Networks for Video Object Detection238
Relation Distillation Networks is presented --- a new architecture that novelly aggregates and propagates object relation to augment object features for detection and achieves superior results when comparing to state-of-the-art methods.
- 229
- 227
- 217
- Hierarchy Parsing for Image Captioning206
A new design to model a hierarchy from instance level, region level and region level to the whole image to delve into a thorough image understanding for captioning and is appealing in view that it is pluggable to any neural captioning models.
- Jointly Localizing and Describing Events for Dense Video Captioning199
This paper presents a novel framework for dense video captioning that unifies the localization of temporal event proposals and sentence generation of each proposal, by jointly training them in an end-to-end manner.
- Learning Spatio-Temporal Representation With Local and Global Diffusion198
A novel framework to boost the spatio-temporal representation learning by Local and Global Diffusion (LGD) is presented, which construct a novel neural network architecture that learns the local and global representations in parallel.
- Multi-Scale Triplet CNN for Person Re-Identification197
A multi-scale triplet convolutional neural network which captures visual appearance of a person at various scales is proposed, addressing the problem of small training set in person re-identification.
- To Create What You Tell180
A novel Temporal GANs conditioning on Captions, namely TGANs-C, in which the input to the generator network is a concatenation of a latent noise vector and caption embedding, and then is transformed into a frame sequence with 3D spatio-temporal convolutions.
- Learning Multimodal Attention LSTM Networks for Video Captioning175
This work presents a novel deep framework to boost video captioning by learning Multimodal Attention Long-Short Term Memory networks (MA-LSTM), and designs a novel child-sum fusion unit in the MA-L STM to effectively combine different encoded modalities to the initial decoding states.
- Dual Vision Transformer174
This paper proposes a novel Transformer architecture that elegantly exploits the global semantics for self-attention learning, namely Dual Vision Transformer (Dual-ViT).
- Boosting image sentiment analysis with visual attention162
Sentiment Networks with visual Attention (SentiNet-A) is presented — a novel architecture that integrates visual attention into the successful Convolutional Neural Networks (CNN) sentiment classification framework, by training them in an end-to-end manner.
- Incorporating Copying Mechanism in Image Captioning for Learning Novel Objects155
A new architecture that incorporates copying into the Convolutional Neural Networks plus Recurrent Neural Networks (RNN) image captioning framework, for describing novel objects in captions, and superior results are reported when compared to state-of-the-art deep models.
- HiDream-I1: A High-Efficient Image Generative Foundation Model with Sparse Diffusion Transformer152
HiDream-I1 is introduced, a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds and evolves to form a comprehensive image agent (HiDream-A1) capable of fully interactive image creation and refinement.
- Customizable Architecture Search for Semantic Segmentation152
A Customizable Architecture Search approach to automatically generate a network architecture for semantic image segmentation that achieves superior performance over several state-of-the-art techniques and is able to search an optimized architecture with customized constraints.
- Comparison of annual energy performances with different ventilation methods for cooling152
Typical configurations of an office, a classroom and a retail shop in Hong Kong are investigated and stratum ventilation derives its energy saving potential largely from the reduction in ventilation and transmission loads and increased COP of chillers.
- Semantic-Conditional Diffusion Networks for Image Captioning*140
The deeply rooted conventions in learning Transformer-based encoder-decoder are broken, and a new diffusion model based paradigm tailored for image captioning, namely Semantic-Conditional Diffusion Networks (SCD-Net) is proposed.
- 134
- Evaluation of thermal comfort conditions in a classroom with three ventilation methods133
It is confirmed that stratum ventilation could provide satisfactory thermal comfort level to rooms of temperature up to 27 °C and the confirmation of the significantly elevated thermal neutral temperature can have a number of implications for both thermal comfort in an air- conditioned room and energy consumption of the associate air-conditioning system.
- Action Recognition by Learning Deep Multi-Granular Spatio-Temporal Video Representation132
This paper presents a novel framework to boost action recognition by learning a deep spatio-temporal video representation at hierarchical multi-granularity using 2D or 3D convolutional neural networks to learn both the spatial and temporal representations.
- Comprehending and Ordering Semantics for Image Captioning128
A new recipe of Transformer-style structure is proposed, namely Comprehending and Ordering Semantics Networks (COS-Net), that novelly unifies an enriched semantic comprehending and a learnable semantic ordering processes into a single architecture.
- SeCo: Exploring Sequence Supervision for Unsupervised Representation Learning126
This paper compositions a trilogy of exploring the basic and generic supervision in the sequence from spatial, spatiotemporal and sequential perspectives, and uniquely regard the signals as the foundation in contrastive learning and derive a particular form named Sequence Contrastive Learning (SeCo).
- 121
- Recurrent Tubelet Proposal and Recognition Networks for Action Detection120
This work presents a novel deep architecture called Recurrent Tubelet Proposal and Recognition (RTPR) networks to incorporate temporal context for action detection and conducts extensive experiments to demonstrate superior results over state-of-the-art methods.
- 113
- 111
- Temporal Deformable Convolutional Encoder-Decoder Networks for Video Captioning109
A novel design --- Temporal Deformable Convolutional Encoder-Decoder Networks (dubbed as TDConvED) that fully employ convolutions in both encoder and decoder networks for video captioning and increases CIDEr-D performance from 58.8% to 67.2% on MSVD.
- Exploring Structure-aware Transformer over Interaction Proposals for Human-Object Interaction Detection104
This work designs a novel Transformer-style HOI detector, i.e., Structure-aware Transformer over Interaction Proposals (STIP), for HOI detection that upgrades vanilla Transformer by encoding additionally the holistically semantic structure among interaction proposals as well as the locally spatial structure of human/object within each interaction proposal, so as to strengthen HOI predictions.
- Mocycle-GAN101
A new Motion-guided Cycle GAN, dubbed as Mocycle-GAN, that novelly integrates motion estimation into unpaired video translator that capitalizes on three types of constrains: adversarial constraint discriminating between synthetic and real frame, cycle consistency encouraging an inverse translation on both frame and motion, and motion translation validating the transfer of motion between consecutive frames.
- 98
- 95
- ObjectFusion: Multi-modal 3D Object Detection with Object-Centric Fusion88
This paper proposes a novel Object-centric Fusion (ObjectFusion) paradigm, which completely gets rid of camera-to-BEV transformation during fusion to align object-centric features across different modalities for 3D object detection.
- Deep Semantic Hashing with Generative Adversarial Networks88
This paper studies the exploration of generating synthetic data through semi-supervised generative adversarial networks (GANs), which leverages largely unlabeled and limited labeled training data to produce highly compelling data with intrinsic invariance and global coherence, for better understanding statistical structures of natural data.
- Exploring Category-Agnostic Clusters for Open-Set Domain Adaptation85
This paper addresses the problem of domain adaptation from closed-set to open-set situation by augmenting the state-of-the-art domain adaptation technique, Self-Ensembling, with category-agnostic clusters in target domain with SE-CC.
- Control3D: Towards Controllable Text-to-3D Generation84
This work presents the first attempt for text-to-3D generation conditioning on the additional hand-drawn sketch, namely Control3D, which enhances controllability for users and can generate accurate and faithful 3D scenes that align closely with the input text prompts and sketches.
- 84
- Click-through-based cross-view learning for image search84
This paper proposes a novel cross-view learning method for image search, named Click-through-based Cross-view Learning (CCL), by jointly minimizing the distance between the mappings of query and image in the latent subspace and preserving the inherent structure in each original space.
- 82
- VideoStudio: Generating Consistent-Content and Multi-scene Videos78
This paper proposes a novel framework, namely VideoStudio, for consistent-content and multi-scene video generation that outperforms the SOTA video generation models in terms of visual quality, content consistency, and user preference.
- Boosting Diffusion Models with Moving Average Sampling in Frequency Domain78
This paper reinterpret the iterative denoising process as model optimization and leverage a moving average mechanism to ensemble all the prior samples and lead to superior performances compared to the baselines, with almost negligible additional complexity cost.
- Pointing Novel Objects in Image Captioning76
This paper presents Long Short-Term Memory with Pointing (LSTM-P) --- a new architecture that facilitates vocabulary expansion and produces novel objects via pointing mechanism by augmenting standard deep captioning architectures with object learners.
- Stand-Alone Inter-Frame Attention in Video Models74
This paper presents a new recipe of inter-frame attention block, namely Stand-alone Inter-Frame Attention (SIFA), that novelly delves into the deformation across frames to estimate local self-attention on each spatial location.
- daBNN74
The proposed daBNN is a super fast inference framework that implements BNNs on ARM devices with several speed-up and memory refinement strategies for bit-packing, binarized convolution, and memory layout to enhance inference efficiency.
- A bibliometric analysis of literature on bibliometrics in recent half-century72
The aim of this study is to understand the current research status and development directions of research on bibliometrics, and implement bibliometric methods based on the Web of Science to analyse the publications, subjects, citation, co-citation, collaboration and keywords of bibliography from 1969 to 2022.
- Deep Quantization: Encoding Convolutional Activations with Deep Generative Model71
This paper presents Fisher Vector encoding with Variational Auto-Encoder (FV-VAE), a novel deep architecture that quantizes the local activations of convolutional layer in a deep generative model, by training them in an end-to-end manner.
- 69
- ControlStyle: Text-Driven Stylized Image Generation Using Diffusion Priors68
This paper presents a new diffusion model (ControlStyle) via upgrading a pre-trained text-to-image model with a trainable modulation network enabling more conditions of text prompts and style images, and introduces diffusion style and content regularizations to facilitate the learning of this modulation network.
- Visualizing and Understanding Patch Interactions in Vision Transformer67
This work proposes a novel explainable visualization approach to analyze and interpret the crucial attention interactions among patches for ViT, and introduces a quantification indicator to measure the impact of patch interaction and verify such quantification on attention window design and indiscriminative patches removal.
- 63
- SD-DiT: Unleashing the Power of Self-Supervised Discrimination in Diffusion Transformer *61
This work novelly unleashes the self-supervised discrimination knowledge to boost DiT training by encoding discriminative pairs with student and teacher DiT encoders, designed to encourage the inter-image alignment in the selfsupervised embedding space.
- 61
- Auto-captions on GIF61
This work presents Auto-captions on GIF (ACTION), which is a new large-scale pre-training dataset for generic video understanding, and provides an evaluation of a Transformer-based encoder-decoder structure for vision-language pre- training, which is further adapted to video captioning downstream task and yields the compelling generalizability on MSR-VTT.
- Scheduled Sampling in Vision-Language Pretraining with Decoupled Encoder-Decoder Network61
This paper proposes a primary scheduled sampling strategy that elegantly mitigates such discrepancy via pretraining encoder-decoder in a two-pass manner and demonstrates the compelling generalizability of the pretrained encoding structure by fine-tuning on four VL understanding and generation downstream tasks.
- Single Shot Video Object Detector60
This paper presents Single Shot Video Object Detector (SSVD) — a new architecture that novelly integrates feature aggregation into a one-stage detector for object detection in videos and takes Feature Pyramid Network (FPN) as backbone network to produce multi-scale features.
- 60
- CARIS: Context-Aware Referring Image Segmentation59
Context-Aware Referring Image Segmentation (CARIS), a novel architecture that enhances the contextual awareness of linguistic features via sequential vision-language attention and learnable prompts, which achieves new state-of-the-art performances on three public benchmarks.
- 59
- Deep Learning for Video Captioning: A Review55
The problem of video captioning is formulated, state-of-the-art methods categorized by their emphasis on vision or language are reviewed, and followed by a summary of standard datasets and representative approaches are reviewed.
- HIRI-ViT: Scaling Vision Transformer With High Resolution Inputs51
HIRI-ViT is built upon the seminal idea of decomposing the typical CNN operations into two parallel CNN branches in a cost-efficient manner and achieves to-date the best published Top-1 accuracy of 84.3% on ImageNet with 448×448 inputs.
- Retrieval Augmented Convolutional Encoder-decoder Networks for Video Captioning50
This article uniquely introduces a Retrieval Augmentation Mechanism (RAM) that enables the explicit reference to existing video-sentence pairs within any encoder-decoder captioning model, and presents Retrival Augmented Convolutional Encoder-Decoder Network (R-ConvED), which novelly integrates RAM into convolutional encoding structure to boost video captioning.
- 50
- A Style and Semantic Memory Mechanism for Domain Generalization49
It is verified that it is critical for the network to be informative on what domain features are invariant and shared among in-stances, so that the network sharpens its understanding and improves its semantic discriminative ability.
- Annotation for free48
It is shown that annotation can be performed for free with competitive performance and minimum computing resource, representing a new and promising paradigm for video tagging in addition to machine learning and data-driven methodologies.
- CoCo-BERT46
It is demonstrated the superiority of CoCo-BERT as a pre-trained structure over a wide range of downstream tasks (e.g., cross-modal retrieval, video question answering, and video captioning), and the CoCo proxy objective can be further integrated into any BERT-type encoder-decoder structure for video-language pre-training.
- Learning Spatial Adaptation and Temporal Coherence in Diffusion Models for Video Super-Resolution44
SATeCo pivots on learning spatial-temporal guidance from low-resolution videos to calibrate both latent-space high-resolution video denoising and pixel-space video reconstruction, and optimizes two deliberately-designed spatial feature adaptation and temporal feature alignment modules in the decoder of UNet and VAE.
- TRIP: Temporal Residual Learning with Image Noise Prior for Image-to-Video Diffusion Models44
TRIP is a new recipe of image-to-video diffusion paradigm that pivots on image noise prior derived from static image to jointly trigger inter-frame relational reasoning and ease the coherent temporal modeling via temporal residual learning.
- Learning a Unified Sample Weighting Network for Object Detection44
A general loss function is devised to cover most region-based object detectors with various sampling strategies, and then a unified sample weighting network is proposed to predict a sample's task weights to consistently boost the detection accuracy.
- Hi3D: Pursuing High-Resolution Image-to-3D Generation with Video Diffusion Models42
High-resolution Image-to-3D model (Hi3D) is presented, a new video diffusion based paradigm that redefines a single image to multi-view images as 3D-aware sequential image generation (i.e., orbital video generation).
- 3D Cascade RCNN: High Quality Object Detection in Point Clouds42
This work presents a simple yet effective cascade architecture that allocates multiple detectors based on the voxelized point clouds in a cascade paradigm, pursuing higher quality 3D object detector progressively, and quantitatively defines the sparsity level of the points within 3D bounding box of each object as the point completeness score.
- MotionPro: A Precise Motion Controller for Image-to-Video Generation*40
This work presents MotionPro, a precise motion controller that novelly leverages region-wise trajectory and motion mask to regulate fine-grained motion synthesis and identify target motion category (i.e., object or camera moving), respectively.
- Modality-Agnostic Debiasing for Single Domain Generalization39
The superiority of the Modality-Agnostic Debiasing framework for single-DG scenarios with different modalities is validated, including recognition on 1D texts, 2D images, 3D point clouds, and semantic segmentation on2D images.
- Motion-Focused Contrastive Learning of Video Representations39
A Motion-focused Contrastive Learning method that capitalizes on optical flow of each frame in a video to temporally and spatially sample the tubelets as data augmentations and aligns gradient maps of the convolutional layers to optical flow maps from spatial, temporal and spatio-temporal perspectives, in order to ground motion information in feature learning.
- Transferring and Regularizing Prediction for Semantic Segmentation39
This paper presents a Regularizer of Prediction Transfer (RPT) that imposes the intrinsic properties of semantic segmentation as constraints to regularize model transfer in an unsupervised fashion.
- 39
- 38
- Convolutional Auto-encoding of Sentence Topics for Image Paragraph Generation38
A new design --- Convolutional Auto-Encoding (CAE) that purely employs convolutional and deconvolutional auto-encoding framework for topic modeling on the region-level features of an image and an architecture, namely CAE plus Long Short-Term Memory (dubbed as CAE-LSTM), that novelly integrates the learnt topics in support of paragraph generation.
- Detecting shot boundary with sparse coding for video summarization38
A new shot boundary detection algorithm, that learns a dictionary from the given video using sparse coding and updates atoms in the dictionary, following the philosophy that different shots cannot be reconstructed using the learned dictionary.
- VP3D: Unleashing 2D Visual Prompt for Text-to-3D Generation37
This work introduces a novel Visual Prompt-guided text-to-3D diffusion model (VP3D) that explicitly unleashes the visual appearance knowledge in 2D visual prompt to boost text-to-3D generation and significantly eases the learning of visual appearance of 3D models and thus leads to higher visual fidelity with more detailed textures.
- X-modaler37
This work proposes X-modaler --- a versatile and high-performance codebase that encapsulates the state-of-the-art cross-modal analytics into several general-purpose stages, and can be simply extended to power startup prototypes for other tasks in cross-Modal analytics, including visual question answering, visual commonsense reasoning, and cross- modal retrieval.
- A Self-Training Method for Machine Reading Comprehension with Soft Evidence Extraction36
A Self-Training method (STM), which supervises the evidence extractor with auto-generated evidence labels in an iterative process, and demonstrates the improvement on existing MRC models.
- Semi-supervised Hashing with Semantic Confidence for Large Scale Visual Search36
This paper proposes a novel semi-supervised hashing framework by leveraging semantic confidence, where a confidence factor is first assigned to each example by neighbor voting and click count in the scenarios with label and click-through data, and incorporated into the pairwise and triplet relationship learning for hashing.
- Lightweight and Progressively-Scalable Networks for Semantic Segmentation34
Extensive experiments conducted on three datasets consistently demonstrate the superiority of LPS-Net over several efficient semantic segmentation methods, and the design of convolutional blocks, and the ways of interactions across multiple scales, all from lightweight standpoint for semantic segmentation are analyzed.
- Representing Videos as Discriminative Sub-graphs for Action Recognition34
A new design of sub-graphs to represent and encode the discriminative patterns of each action in the videos of Something-Something V1 & V2 and Kinetics-400 datasets are introduced, and superior results are reported when comparing to state-of-the-art methods.
- Human Gait Recognition Based on Multiple Feature Combination and Parameter Optimization Algorithms34
An improved gait recognition algorithm, which combines multiple feature combination and artificial bee colony for optimizing the support vector machine (ABC-SVM), and the combined feature samples not only improve the gait classification accuracy but also enhance the recognition stability.
- Boosting Scene Graph Generation with Visual Relation Saliency33
This work presents a new design, named as Saliency-guided Message Passing (SMP), that boosts the generation of such scene graph structure with the guidance from the visual relation saliency, and empirically validate the compelling generalizability of the learned scene graphs via SMP on downstream tasks like cross-model retrieval and image captioning.
- 33
- A molecular atlas reveals the tri-sectional spinning mechanism of spider dragline silk32
A systematic molecular atlas of dragline silk production through a high-quality genome assembly for the golden orb-weaving spider Trichonephila clavata and a multiomics approach to defining the Ma gland tri-sectional architecture is performed, revealing the single-cell spatial architecture of silk production in the Ma glands.
- Dynamic Temporal Filtering in Video Models32
This paper presents a new recipe of temporal feature learning, namely Dynamic Temporal Filter (DTF), that novelly performs spatial-aware temporal modeling in frequency domain with large temporal receptive field and dynamically learns a specialized frequency filter for every spatial location to model its long-range temporal dynamics.
- 30
- Transferrable Contrastive Learning for Visual Domain Adaptation30
This work presents a particular paradigm of self-supervised learning tailored for domain adaptation, i.e., Transferrable Contrastive Learning (TCL), which links the SSL and the desired cross-domain transferability congruently, and finds contrastive learning intrinsically a suitable candidate fordomain adaptation.
- Deep Domain Adaptation Hashing with Adversarial Learning30
A novel Deep Domain Adaptation Hashing with Adversarial learning (DeDAHA) architecture is presented, which mainly consists of three components: a deep convolutional neural networks for learning basic image/frame representation followed by an adversary stream on one hand to optimize the domain discriminator, and on the other, to interact with each domain-specific hashing stream for encoding image representation to hash codes.
- Multi-feature gait recognition with DNN based on sEMG signals28
A gait recognition method based on the deep neural network of surface electromyography (sEMG) signals to improve the stability and accuracy of gait Recognition using sEMG signals of the lower limbs is proposed.
- $\mathbb {Z}_{p}\mathbb {Z}_{p^{s}}$-additive cyclic codes are asymptotically good28
It is proved that p-additive cyclic codes are asymptotically good since the asymPTotic GV-bound at 1+p^{s-1}}{2}\delta is greater than 1 and the relative distance of the code is convergent to δ, while the rate is converge to 1.
- Click-boosting multi-modality graph-based reranking for image search28
The use of click-through data, which can be viewed as the footprints of user searching behavior, is explored as an effective means of understanding query, for providing the basis on identifying the recurrent patterns that are potentially helpful for reranking.
- 3DStyle-Diffusion: Pursuing Fine-grained Text-driven 3D Stylization with 2D Diffusion Models27
This work proposes a new 3DStyle-Diffusion model that triggers fine-grained stylization of 3D meshes with additional controllable appearance and geometric guidance from 2D Diffusion models, and elegantly integrates both image rendering via implicit MLP networks and diffusion process of image synthesis in an end-to-end fashion.
- Uni-EDEN: Universal Encoder-Decoder Network by Multi-Granular Vision-Language Pre-training27
A pre-trainable Universal Encoder-DEcoder Network (Uni-EDEN), consisting of three modules: object and sentence encoders that separately learns the representations of each modality and sentence decoder that enables both multi- modal reasoning and sentence generation via inter-modal interaction.
- 26
- Lévy noise-induced transition and stochastic resonance in a tumor growth model26
The results indicate that both Levy and Gaussian noise sources give rise to noise-induced transitions for the system, with a peculiarity that smaller stability indexα and Levy noise intensity D enhance the likelihood of tumor cell death.
- ReCO: A Large Scale Chinese Reading Comprehension Dataset on Opinion26
Quality analysis demonstrates the challenge of ReCO that it requires various types of reasoning skills such as causal inference, logical reasoning, etc to be a good challenge for machine reading comprehension.
- Long Short-Term Relation Networks for Video Action Detection26
A new Long Short-Term Relation Networks is presented, dubbed as LSTR, that novelly aggregates and propagates relation to augment features for video action detection and superior results are reported when comparing to state-of-the-art methods.
- PointClustering: Unsupervised Point Cloud Pre-training using Transformation Invariance in Clustering25
- Smart Director: An Event-Driven Directing System for Live Broadcasting25
The Smart Director system is the first end-to-end automated directing system for multi-camera sports broadcasting, completely driven by the semantic understanding of sports events, and is also the first system to solve the novel problem of multi-view joint event detection by cross-view relation modeling.
- Asymptotically good Z p r Z p s -additive cyclic codes25
It is proved that there exist numerous asymptotically good Z p r Z p s -additive cyclic codes.
- 25
- Skew cyclic codes over $F_{q}+uF_{q}+vF_{q}+uvF_{q}$25
The structural properties of skew cyclic codes over the ring R are investigated through a decomposition theorem and a formula for the number of skewcyclic codes of length $n$ over $R$ is given.
- HiDream-I1: An Open-Source High-Efficient Image Generative Foundation Model24
HiDream-I1, a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds, and HiDream-E1, a new image editing model namely HiDream-E1 that quickly gained strong traction in the community.
- 24
- 24
- 23
- Knowledge Transfer in Incremental Learning for Multilingual Neural Machine Translation23
Experimental results show that the proposed knowledge transfer method can learn new knowledge from diverse language pairs incrementally meanwhile maintaining performance on original language pairs, outperforming various strong baselines in incremental learning for MNMT.
- Visual Autoregressive Modeling for Instruction-Guided Image Editing22
This paper presents VAREdit, a visual autoregressive (VAR) framework that reframes image editing as a next-scale prediction problem and introduces a Scale-Aligned Reference (SAR) module, which injects scale-matched conditioning information into the first self-attention layer.
- A Low Rank Promoting Prior for Unsupervised Contrastive Learning22
A novel probabilistic graphical model that effectively incorporates the low rank promoting prior into the framework of contrastive learning, referred to as LORAC, which clearly surpasses the state-of-the-art approaches on multiple benchmarks, including image classification, object detection, instance segmentation and keypoint detection.
- Deep Learning–Based Multimedia Analytics22
The development path of major tasks in multimedia analytics is reviewed and the performance review of popular benchmarks gives a pathway to technology advancement and helps identify both milestone works and future directions.
- Transforming Radiance Field with Lipschitz Network for Photorealistic 3D Scene Stylization21
A concise and flexible learning framework namely LipRF is built, which upgrades arbitrary 2D PST methods with Lipschitz mapping tailored for the 3D scene, and devise an adaptive regularization to balance the reconstruction and stylization.
- DPTDR: Deep Prompt Tuning for Dense Passage Retrieval21
The experimental results show that the proposed method, called DPTDR, outperforms previous state-of-the-art models on both MS-MARCO and Natural Questions and facilitates the industry, as it saves enormous efforts and costs of deployment and increases the utility of computing resources.
- Context-based friend suggestion in online photo-sharing community21
This work proposes a new approach, named context-based friend suggestion, to leverage the diverse form of contextual cues for more effective friend suggestion in the social media community, and develops two user-based similarity measurements, i.e., visual similarity and geo similarity for characterizing user relationship.
- Improving Text-Guided Object Inpainting with Semantic Pre-inpainting20
This paper proposes to decompose the typical single-stage object inpainting into two cascaded processes: 1) semantic pre-inpainting that infers the semantic features of desired objects in a multi-modal feature space; 2) high-fieldity object generation in diffusion latent space that pivots on such inpainted semantic features.
- 20
- Learning hierarchical video representation for action recognition20
This work represents the hierarchical structure of video with multiple granularities including, from short to long, single frame, consecutive frames (motion), short clip, and the entire video and proposes a novel deep learning framework to model each granularity individually.
- Predicting domain adaptivity20
Different parameters, including shift of data distribution, visual and context diversities, that may hinder or otherwise encourage the recycling of old classifiers for new dataset are explored and empirical insights of when to recycle available resources are given.
- MLP-3D: A MLP-like 3D Architecture with Grouped Time Mixing19
MLP-3D networks are presented, a novel MLP-like 3D architecture for video recognition without the dependence on convolutions or attention mechanisms, and achieves 68.5%/81.4% top-1 accuracy on Something-Something V2 and Kinetics-400 datasets, respectively.
- Document Gated Reader for Open-Domain Question Answering19
A document gated reader to generate the right answer from multiple documents is proposed with a document-level gate operation to determine the question-document relevance and embed it into the answer generation process, and optimize it with the global normalization objective.
- Click-through-based Subspace Learning for Image Search18
It is demonstrated that the above two fundamental challenges can be mitigated by jointly exploring the subspace learning and the use of click-through data and the former aims to create a latent subspace with the ability in comparing information from the original incomparable views.
- 17
- Learning to Localize Actions from Moments17
This paper introduces a new design of transfer learning type to learn action localization for a large set of action categories, but only on action moments from the categories of interest and temporal annotations of untrimmed videos from a small set of Action Herald Networks (AherNet).
- Deep Metric Learning With Density Adaptivity17
This paper integrates density, i.e., the measure of data concentration in the representation, into the optimization of DML frameworks to adaptively balance inter-class similarity and intra-class variation by training the architecture in an end-to-end manner.
- Interactive Conversational Head Generation16
A new conversation head generation benchmark for synthesizing behaviors of a single interlocutor in a face-to-face conversation and proposes corresponding baseline solutions to three novel tasks targeting the interaction modeling during the face-to-face conversation.
- Share-and-Chat16
This paper presents a novel two-stage deep learning-based approach to automatic video commenting that outperforms several state-of-the-art methods and achieves human-level video commenting.
- Image search by graph-based label propagation with image representation from DNN16
This paper proposes a graph-based label propagation algorithm that employs neighborhood graph search to find the nearest neighbors on an image similarity graph built up with visual representations from deep neural networks and further aggregates their clicked queries/click counts to get the labels of the new image.
- 15
- Bottom-up and Top-down Object Inference Networks for Image Captioning15
This work presents the Bottom-up and Top-down Object inference Network (BTO-Net), which novelly exploits the object sequence of interest as top-down signals to guide image captioning and obtains competitive performances on the COCO benchmark.
- CAPO: Towards Enhancing LLM Reasoning through Generative Credit Assignment14
CAPO can help the model to foster the learning of correct reasoning pathways, encouraging the generation of trajectories conducive to correct results and consistently outperforms supervised learning-based and RL-based fine-tuning methods across six challenging mathematical benchmarks and three out-of-domain benchmarks.
- 14
- Condensing a Sequence to One Informative Frame for Video Recognition14
A novel Informative Frame Synthesis (IFS) architecture that incorporates three objective tasks, i.e., appearance reconstruction, video categorization, motion estimation, and two regularizers, and achieves comparable performance with the state-of-the-art methods with less computational cost.
- Unified entity search in social media community14
A unified, multi-level, and correlative entity graph to represent the unstructured social media data, through which various applications can be realized more effectively in one single framework is investigated.
- Pursuing Temporal-Consistent Video Virtual Try-On via Dynamic Pose Interaction13
A new framework, namely Dynamic Pose Interaction Diffusion Models (DPIDM), to leverage diffusion models to delve into dynamic pose interactions for video virtual try-on and capitalizes on a temporal regularized attention loss between consecutive frames to enhance temporal consistency.
- Incorporating Visual Correspondence into Diffusion Model for Virtual Try-On13
This work proposes to explicitly capitalize on visual correspondence as the prior to tame diffusion process instead of simply feeding the whole garment into UNet as the appearance reference, and interprets the fine-grained appearance and texture details as a set of structured semantic points, which are augmented into 3D-aware cues with depth/normal map of target person.
- Unleashing Text-to-Image Diffusion Prior for Zero-Shot Image Captioning13
A novel Patch-wise Cross-modal feature Mix-up (PCM) mechanism to adaptively mitigate the unfaithful contents in a fine-grained manner during training, which can be integrated into most of encoder-decoder frameworks, and ranks first in both in-domain and cross-domain zero-shot image captioning.
- Deep Equilibrium Multimodal Fusion13
This paper proposes a novel deep equilibrium (DEQ) method towards multimodal fusion via seeking a fixed point of the dynamic multi-modality fusion process and modeling the feature correlations in an adaptive and recursive manner.
- 13
- 13
- 13
- Contextual and selective attention networks for image captioning12
A contextual and selective attention network that novelly memorizes contextual attention and brings out the principal components from each attention is presented, and extensive experiments on the COCO image captioning dataset demonstrate the superiority of CoSA-Net.
- Nonbinary quantum codes from constacyclic codes over polynomial residue rings12
It is proved that the image of a (1-u) -constacyclic code of length n over R under the Gray map is a distance-invariant linear cyclic codes of length pn over F q 2.
- Skew Cyclic Codes over a Non‐chain Ring12
The generator polynomial of the dual code of a skew cyclic code are obtained and the structural properties of skewcyclic codes over the ring by a decomposition theorem are investigated.
- 11
- FreeEnhance: Tuning-Free Image Enhancement via Content-Consistent Noising-and-Denoising Process11
This paper proposes a novel framework, namely FreeEnhance, for content-consistent image enhancement using the off-the-shelf image diffusion models, and outperforms the state-of-the-art image enhancement models in terms of quantitative metrics and human preference.
- On Self‐dual and LCD Double Circulant Codes over a Non‐chain Ring*11
When n is an odd prime, using random coding, the obtained families of asymptotically good self-dual and LCD codes of length 6n over 𝔽 q are obtained.
- VIREO @ TRECVID 2014: Instance Search and Semantic Indexing11
This paper summarizes the following two tasks participated by VIREO group: instance search and semantic indexing and presents the approaches and analyze the results obtained in TRECVID 2014 benchmark evaluation.
- Bi-calibration Networks for Weakly-Supervised Video Representation Learning10
A new design of mutual calibration between query and text to achieve more reliable visual-textual supervision for video representation learning is introduced and Bi-Calibration Networks (BCN) is presented that novelly couples two calibrations to learn the correction from text to query and vice versa.
- Boosting Vision-and-Language Navigation with Direction Guiding and Backtracing10
A Direction-guided Navigator Agent (DNA) that novelly integrates direction clues derived from instructions into the essential encoder-decoder navigation framework and strengthens the alignment between instruction and action sequence.
- 10
- Vision and language: from visual perception to content creation10
This paper reviews the recent advances along these two dimensions: “vision to language” and “language to vision,” which mainly focuses on the development of image/video captioning, as well as typical encoder–decoder structures and benchmarks, while the latter summarizes the technologies of visual content creation.
- Seeing Bot10
A video captioning bot, named Seeing Bot, which can generate a natural language description about what it is seeing in near real time and sends the speech to an end user via an earphone.
- Skew Cyclic Codes over $\mathbb{F}_{q}+v\mathbb{F}_{q}+v^{2}\mathbb{F}_{q}$10
The structural properties of skew cyclic codes over ring $R$ are investigated by a decomposition theorem and the generator polynomials of the duals of skewcyclic codes are described.
- 9
- SPE-Net: Boosting Point Cloud Analysis via Rotation Robustness Enhancement9
Intriguingly, it is revealed that by locally encoding the rotation information through SPE-Net, the rotation-invariant features are still of critical importance in benefiting the test samples without any actual global rotation.
- Boosting Video Representation Learning with Multi-Faceted Integration9
A new learning framework to aggregate facets from different datasets for learning a representation that could reflect the full spectrum of video content, which explicitly maps video representation into a rich semantic embedding space, and jointly optimizes video representation from two perspectives.
- Learning Click-Based Deep Structure-Preserving Embeddings with Visual Attention9
A novel click-based Deep Structure-Preserving Embeddings with visual Attention (DSPEA) model is presented, which consists of two components: deep convolutional neural networks followed by image embedding layers for learning visual embedding, and a deep neural networks for generating query semantic embedding.
- MSR-VTT: A Large Video Description Dataset for Bridging Video and Language [Supplementary Material]9
When organizing the Microsoft Research Video To Language challenge, some sentences annotated by AMT workers are identical in one video clip or very similar in one category so the performance cannot be well matched with what was reported in the CVPR paper.
- 8
- 8
- 8
- Core-Text: Improving Scene Text Detection with Contrastive Relational Reasoning8
This work quantitatively analyze the sub-text problem and presents a simple yet effective design, COntrastive RElation (CORE) module, to mitigate that issue and integrates the CORE module into a two-stage text detector of Mask R-CNN and devise the text detector CORE-Text.
- 8
- Animating Your Life8
An unified unpaired video-to-video translator is utilized to explore both appearance structure and temporal continuity in video synthesis, which can transform real-world video into cartoon or ink-wash animation in real-time.
- 8
- Incorporating seasonal time series analysis with search behavior information in sales forecasting8
An algorithm using click-through and query reformulation information to cluster related queries and count their frequencies on monthly-basis is designed, and Exponentially Weighted Moving Averages (EWMA) model is introduced to measure the seasonal impact on the sales trend.
- 7
- 7
- 7
- 7
- Multi-Lingual Question Generation with Language Agnostic Language Model7
A language-agnostic language model is developed, which learns the shared representation from several languages in a single architecture in a single architecture to improve the multi-lingual question generation via pre-training.
- 7
- 7
- 7
- 7
- Video ChatBot7
A video chatbot that can generate human-level emotional comments referring to the videos shared by users and trigger a conversation with users is demonstrated, which has great potential to increase live social interactions.
- VIREO@TRECVID 2012: Searching with Topology, Recounting will Small Concepts, Learning with Free Examples7
The vireo group participated in four tasks: instance search, multimediaevent recounting, multimedia event detection, and semantic indexing, and the approaches and evaluation results are presented and discussed.
- Selective Volume Mixup for Video Action Recognition6
A novel video augmentation strategy named Selective Volume Mixup (SV-Mix) is proposed to improve the generalization ability of deep models with limited training videos and empirically demonstrate the merits of the SV-Mix augmentation on a wide range of video action recognition benchmarks.
- 6
- 6
- 6
- 6
- Learning and Evaluating Human Preferences for Conversational Head Generation6
Experimental results validate the superiority of Preference Score in aligning with human perception, and also demonstrate robustness and generalizability to unseen data, making it a valuable tool for advancing conversation head generation.
- 6
- ComQA: Compositional Question Answering via Hierarchical Graph Neural Networks6
A hierarchical graph neural networks is proposed, which represent the document from the low-level word to the high-level sentence, which achieves a significant improvement over previous machine reading comprehension methods and pre-training methods.
- Cycle or Minkowski6
This study points out the implications of Minkowski metric used in KG embedding and introduces a new metric, named Cycle metric, based on the oscillation property of the periodic function, which shows that the function period has a significant influence on the expressiveness of the embedding space.
- A Family of Constacyclic Codes over 𝔽 2 m + u 𝔽 2 m and Its Application to Quantum Codes6
It is proved that every code of length 2mn which is the Gray image of cyclic codes over R of length n is permutation equivalent to a binary quasi-cyclic code of index m.
- 6
- Double cyclic codes over 𝔽q + u𝔽q + u2𝔽q6
A double cyclic code of length r, s over a chain ring R is a set that can be partitioned into two parts that any cyclic shift of the coordinates of both parts leaves invariant the code.
- Click-boosting random walk for image search reranking6
The use of click-through data, which can be viewed as the footprints of user searching behavior, is explored as an effective means of understanding query, for providing the basis on identifying the recurrent patterns that are potentially helpful for reranking.
- 6
- 6
- VTON-VLLM: Aligning Virtual Try-On Models with Human Preferences5
A dataset of synthesized VTON images annotated with human judgments across multiple perceptual criteria is curated, and a vision large language model (VLLM), namely VTON-VLLM, is learnt on these annotations, which functions as a unified “fashion expert” and is capable of both evaluating and steering VTON synthesis towards human preferences.
- From Faithfulness to Correctness: Generative Reward Models that Think Critically5
The Thinking-supervised Reward Model (TRM), which incorporates sentence-level thinking supervision to endow reward models with critical thinking abilities, is proposed, which substantially improves the identification of incorrect sentences and incorporates TRM into policy optimization leads to significant gains in both answer correctness and usefulness.
- 5
- 5
- Explaining Cross-domain Recognition with Interpretable Deep Classifier5
A novel Interpretable Deep Classifier that learns the nearest source samples of a target sample as evidence upon which the classifier makes the decision, which leads to a more explainable model with almost no accuracy degradation and effectively calibrates classification for optimum reject options.
- ChiQA5
A new question answering dataset based on image-ChiQA, which contains the real-world queries issued by internet users, combined with several related open-domain images, shows ChiQA requires a deep understanding of both language and vision, including grounding, comparisons, and reading.
- 5
- Double cyclic codes over q + u q + u 2 q .5
A double cyclic code of length r, s over a chain ring R is a set that can be partitioned into two parts that any cyclic shift of the coordinates of both parts leaves invariant the code.
- 5
- 5
- A novel scheme of ROI detection and transcoding for mobile devices in high-definition videoconferencing5
A novel scheme of ROI detection and transcoding for mobile devices in video-conferencing system that uses a Kinect based approach to detect ROIs and a ROI-enabled transcoding scheme will be used to transcode the video streams by using a re-quantization approach.
- Creatively Upscaling Images with Global-Regional Priors4
This work presents C-Upscale, a new recipe of tuning-free image upscaling that pivots on global-regional priors derived from given global prompt and estimated regional prompts via Multimodal LLM to generate ultra-high-resolution images with higher visual fidelity and more creative regional details.
- Identity-Preserving Video Generation Challenge4
The Identity-Preserving Video Generation (IPVG) grand challenge is introduced, a novel dataset comprising approximately 500,000 video-prompt pairs with 200,000 unique identities, each coupled with a reference identity image is released.
- 4
- 4
- Bidirectional Knowledge Reconfiguration for Lightweight Point Cloud Analysis4
This paper explores feature distillation for lightweight point cloud models and proposes a feature mover's distance (FMD) loss based on optimal transportation, which can measure the distance between unordered point cloud features effectively.
- Silver-Bullet-3D at ManiSkill 2021: Learning-from-Demonstrations and Heuristic Rule-based Methods for Object Manipulation4
An overview and comparative analysis of the systems designed for the following two tracks in SAPIEN ManiSkill Challenge 2021: No Interaction Track and No Restriction Track are presented.
- Learning More with Less: A Dynamic Dual-Level Down-Sampling Framework for Efficient Policy Optimization3
This work proposes the Dynamic Dual-Level Down-Sampling framework, a framework that prioritizes the most informative samples and tokens across groups to improve the efficient of policy optimization, and employs a dynamic down-sampling schedule inspired by curriculum learning.
- G-Core: A Simple, Scalable and Balanced RLHF Trainer3
G-Core introduces a parallel controller programming model, enabling flexible and efficient orchestration of complex RLHF workflows without the bottlenecks of a single centralized controller, and proposes a dynamic placement schema that adaptively partitions resources and schedules workloads.
- 3
- Out-of-Distribution Detection with Hilbert-Schmidt Independence Optimization3
This paper proposes an alternative probabilistic paradigm that is both practically useful and theoretically viable for the OOD detection tasks and imposes statistical independence between inlier and outlier data during training, in order to ensure that inlier data reveals little information about OOD data to the deep estimator during training.
- 3
- iDirector: An Intelligent Directing System for Live Broadcast3
iDirector is demonstrated, which aims to mimic the human-in-loop live broadcasting process by aggregating the input multi-camera video streaming into the final output program video (PGM video) for audience.
- 3
- See and chat: automatically generating viewer-level comments on images3
This work proposes a novel two-stage approach, consisting of similar image search and comment ranking, which achieves superior performance and can suggest viewer-level comments.
- 3
- VIREO @ TRECVID 2011: Instance Search, Semantic Indexing, Multimedia Event Detection and Known-Item Search.3
The vireo group participated in four tasks: instance search, semantic indexing, multimedia event detection and known-item search, and the approaches and evaluation results are presented and discussed.
- Investigating Characteristics of Non-click Behavior Using Query Logs3
This paper investigates non-click behavior using large scale search logs from a commercial search engine and analyzes query and non- click behavior characteristics on three levels, i.e., query, session and user level.
- 2
- 2
- Edit-by-Example: Adaptive Exemplar-Based Image Editing2
An exemplar-based image editing framework named Edit-by-Example, which eliminates the reliance on textual prompts, and requires only a single pair of before-and-after images to encapsulate the desired editing effect that can readily be applied on the user-provided query image without any model fine-tuning.
- Visual-Aware Text-to-Speech*2
A new visual-aware text-to-speech (VA-TTS) task to synthesize speech conditioned on both textual inputs and sequential visual feedback of the listener in face- to-face communication and devise a baseline model to fuse phoneme linguistic information and listener visual signals for speech synthesis.
- Temporal and spatial changes in microbial communities and their correlation with volatile aroma compounds during the fermentation of Mucor racemosus sufu2
The results revealed that Proteobacteria and Actinobacteria were the dominant bacterial phyla in the final fermented product and have important implications for isolating specific strains, enhancing flavor, and providing a solid foundation for optimizing and ensuring the quality of sufu.
- 2
- Boosting Relationship Detection in Images with Multi-Granular Self-Supervised Learning2
This article integrates the standard relationship detection objective with a series of proposed self-supervised proxy tasks, which is named as Multi-Granular Self-Supervised learning (MGS), appealing in view that it is pluggable to any neural relationship detection models by simply including the proxy tasks during training, without increasing the computational cost at inference.
- CoCo-BERT: Improving Video-Language Pre-training with Contrastive Cross-modal Matching and Denoising2
- VireoJD-MM @ TRECVid 2019: Activities in Extended Video (ActEV).2
The system developed for Activities in Extended Video(ActEV) task at TRECVid 2019 and the achieved results are described and the system is extended for two aspects separately: better object detection and advanced two-stream action classification.
- Unsupervised Story Comprehension with Hierarchical Encoder-Decoder2
This paper proposes an unsupervised sequence-to-sequence method for story reading comprehension that only adopt the unlabeled story and directly model the context-target inference probability, and proposes a loss-reweight training strategy for the seq- to-seq model to dynamically tuning the training process.
- 2
- Human Behavior Understanding2
This tutorial will present recent advances under the umbrella of human behaviour understanding, which range from the fundamental problem of how to learn "good" video representations, to the challenges of categorizing video content into human action classes, finally to multimedia event detection and surveillance event detection in complex scenarios.
- 2
- VIREO/ECNU @ TRECVID 2013: A Video Dance of Detection, Recounting and Search with Motion Relativity and Concept Learning from Wild2
The VIREO group participated in four tasks: instance search, multimedia event recounting, multimedia event detection, multimedia event detection, and semantic indexing, and this paper will present the approaches and discuss the evaluation results.
- Talk, Imagine, Evolve: A Unified Multimodal Agent for Seamless Visual Generation and Editing1
This paper demonstrates a pioneering unified multimodal agent that transforms complex visual content creation into an intuitive, conversational experience, allowing users to talk, imagine, and evolve their ideas within a single AI interface.
- 1
- Registry for vascular cognitive impairment treatment with traditional Chinese medicine (REVIEW-TCM): Rationale and design of a prospective, observational study1
The REVIEW-TCM study is the first comprehensive, prospective, mixed-methods, registry-based study to evaluate TCM treatment in VCI, which will analyze the effectiveness and safety of TCM in the real world and explore population characteristics and subtypes of VCI suitable for TCM.
- Bridging the Gap Between Training and Inference of Bayesian Controllable Language Models1
A Gemini Discriminator for controllable language generation is proposed which alleviates the mismatch problem with a small computational cost and achieved new state-of-the-art results in automatic and human evaluations.
- 1
- 1
- Adversarial Examples with Difficult Common Words for Paraphrase Identification1
A novel algorithm is proposed to generate a new type of adversarial examples to study the robustness of deep paraphrase identification models and it is shown that adversarial training with generated adversarialExamples can improve model robustness.
- 1
- 1
- Algorithm of optical remote sensing image registration based on strong edge region1
Aiming at the registration of optical remote sensing images, an algorithm based on strong edge region is proposed that combines with the regions' moment invariants and RANSAC method to obtain an accurate match of the strong edge regions.
- Motion Controlled Text-to-4D Generation with Multi-View Video Diffusion Model–
Motion4D is proposed, a rigging-free framework that animates static humanoid meshes with arbitrary initial poses by replicating the motion pattern of a reference sequence by adopting a generate-then-animate paradigm and introducing a Gaussian-Proxied Mesh Animation algorithm.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Multimodal Content Creation, Consumption and Distribution–
This talk will first delve into the core technological breakthroughs in multimodal content generation, particularly the latest advancements in image and video generation tasks, and present an agent-based solution from HiDream.ai to interpret user intention and manage content creation-from ideation to final distribution-using three core, interconnected agents.
- –
- Scale Attention for Learning Deep Face Representation: A Study Against Visual Scale Variation–
This work reform the conv layer by resorting to the scale-space theory, and achieves two-fold facilities, which accomplish the hierarchical scale attention by stacking the reformed layers, building a novel style named SCAN-CNN.
- ChiQA: A Large Scale Image-based Real-World Question Answering Dataset for Multi-Modal Understanding–
- –
- Triple Cyclic Codes Over 𝔽q + u𝔽q–
The generator polynomials and the minimum generating sets of this kind of codes are studied and some optimal or almost optimal linear codes are obtained from this family of codes.
- –
- –
- –
- –
- –
- –
- A family of constacyclic codes over $\mathbb{F}_{2^{m}}+u\mathbb{F}_{2^{m}}$ and application to quantum codes–
It is proved that every code of length $2mn$ which is the Gray image of cyclic codes over $n$ is permutation equivalent to a binary quasi-cyclic code of index $m.
- –
- –
- Macroblock Importance Based Method for Stereo Video Congestion Control–
A stereo video congestion control method, based on the macroblock importance, is proposed, where every macroblock is calculated according to its influence on the current frame and future frames and all macroblocks of a frame are divided into three levels.
- –
- –
- A Hybrid Discriminative Approach with Bayesian Prior Constraint–
The proposed hybrid discriminative approach with Bayesian prior constraints can effectively update the model by figuring out the discriminating samples and correct the misclassifications caused by the uneven distribution of limited samples, and can improve the performances of scene categorization.
- ROI-based Video Quality Assessment and Regulation for Mobile Videoconferencing–
The experimental results show that the adaption scheme can effectively perform bitrate allocation of ROI video streams to make sure that mobile video viewers can perceive the optima video quality based on constrained network connections.
- –
- –
- Task-role-based access control in manufacturing enterprise content management system–
A dynamic and static method of access control based on combining role management with workflow management of manufacturing enterprise content management system is put forward, in which roles and tasks put in the same layer.
- –
- Multi-Model Supported Real-time Character Animation Library–
The paper mainly discusses the Multi-model Supported Real-time Character Animation Library, including the system architecture, the character script and the animation engine, and the introduction of Locomotion Definite State Machine.
- Hand-gesture-based Virtual Studio–
A novel virtual reality system, hand-gesture-based virtual studio was provided and it is the best trait of it that the system provides user strong immersive experience.
- Self-adaptive fusion method for remote sensing images based on regional performance–
In order to consider image fusion evaluation information when choose a fusion method, a self-adaptive fusion method for SAR and optical images based on regional performance was proposed and can obtain satisfactory effect.
- Developing the Computation Software for Cartographic Mathematical Basis–
The development and integration of related softwares for compuatation of map mathematical base are of great importance in the practical production.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-15. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.