Ruicheng Feng lists you as their PhD advisor. Claim this profile to confirm it and keep the rest of your record right.
Claim this profileAcademic lineage
View as a treeStudents and postdocs3
Possible advisorsa guess from early papers, not confirmed
- Shaogang GongSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works411 from public data
- Image Super-Resolution Using Deep Convolutional Networks10,021
This work proposes a deep learning method for single image super-resolution (SR) that directly learns an end-to-end mapping between the low/high-resolution images, represented as a deep convolutional neural network (CNN) that takes the low-resolution image as the input and outputs the high-resolution one.
- Learning a Deep Convolutional Network for Image Super-Resolution5,839
This work proposes a deep learning method for single image super-resolution (SR) that directly learns an end-to-end mapping between the low/high-resolution images and shows that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network.
- ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks4,943
This work thoroughly study three key components of SRGAN – network architecture, adversarial loss and perceptual loss, and improves each of them to derive an Enhanced SRGAN (ESRGAN), which achieves consistently better visual quality with more realistic and natural textures than SRGAN.
- Learning to Prompt for Vision-Language Models4,550
Context Optimization (CoOp) is proposed, a simple approach specifically for adapting CLIP-like vision-language models for downstream image recognition that achieves superb domain generalization performance compared with the zero-shot model using hand-crafted prompts.
- Accelerating the Super-Resolution Convolutional Neural Network3,558
This paper aims at accelerating the current SRCNN, and proposes a compact hourglass-shape CNN structure for faster and better SR, and presents the parameter settings that can achieve real-time performance on a generic CPU while still maintaining good performance.
- Conditional Prompt Learning for Vision-Language Models2,499
Conditional Context Optimization (CoCoOp), which extends CoOp by further learning a lightweight neural network to generate for each image an input-conditional token (vector), and yields stronger domain generalization performance as well.
- Zero-Reference Deep Curve Estimation for Low-Light Image Enhancement2,451
A novel method, Zero-Reference Deep Curve Estimation (Zero-DCE), which formulates light enhancement as a task of image-specific curve estimation with a deep network and shows that it generalizes well to diverse lighting conditions.
- 1,955
- WIDER FACE: A Face Detection Benchmark1,837
There is a gap between current face detection performance and the real world requirements, and the WIDER FACE dataset, which is 10 times larger than existing datasets is introduced, which contains rich annotations, including occlusions, poses, event categories, and face bounding boxes.
- Domain Generalization: A Survey1,753
For the first time, a comprehensive literature review in DG is provided to summarize the developments over the past decade and conduct a thorough review into existing methods and theories.
- Hybrid Task Cascade for Instance Segmentation1,565
This work proposes a new framework, Hybrid Task Cascade (HTC), which differs in two important aspects: (1) instead of performing cascaded refinement on these two tasks separately, it interweaves them for a joint multi-stage processing; (2) it adopts a fully convolutional branch to provide spatial context, which can help distinguishing hard foreground from cluttered background.
- Facial Landmark Detection by Deep Multi-task Learning1,547
A novel tasks-constrained deep model is formulated, with task-wise early stopping to facilitate learning convergence and reduces model complexity drastically compared to the state-of-the-art method based on cascaded deep model.
- 1,456
- EDVR: Video Restoration With Enhanced Deformable Convolutional Networks1,341
This work proposes a novel Video Restoration framework with Enhanced Deformable convolutions, termed EDVR, and proposes a Temporal and Spatial Attention (TSA) fusion module, in which attention is applied both temporally and spatially, so as to emphasize important features for subsequent restoration.
- Recovering Realistic Texture in Image Super-Resolution by Deep Spatial Feature Transform1,224
It is shown that it is possible to recover textures faithful to semantic classes in a single network conditioned on semantic segmentation probability maps through a novel Spatial Feature Transform (SFT) layer that generates affine transformation parameters for spatial-wise feature modulation.
- PSANet: Point-wise Spatial Attention Network for Scene Parsing1,191
The point-wise spatial attention network (PSANet) is proposed to relax the local neighborhood constraint and achieves top performance on various competitive scene parsing datasets, including ADE20K, PASCAL VOC 2012 and Cityscapes, demonstrating its effectiveness and generality.
- 1,102
- CARAFE: Content-Aware ReAssembly of FEatures1,010
Comprehensive evaluations on standard benchmarks in object detection, instance/semantic segmentation and inpainting show CARAFE shows consistent and substantial gains across all the tasks, and has great potential to serve as a strong building block for future research.
- Compression Artifacts Reduction by a Deep Convolutional Network894
A compact and efficient network for seamless attenuation of different compression artifacts is formulated and it is demonstrated that a deeper model can be effectively trained with the features learned in a shallow network.
- A large-scale car dataset for fine-grained categorization and verification882
This paper presents an on-going effort in collecting a large-scale dataset, “CompCars”, that covers not only different car views, but also their different internal and external parts, and rich attributes, and demonstrates a few important applications exploiting the dataset.
- Extract Free Dense Labels from CLIP856
The finding suggests that MaskCLIP can serve as a new reliable source of supervision for dense prediction tasks to achieve annotation-free segmentation in pixel-level dense prediction, specifically in semantic segmentation.
- Learning Lightweight Lane Detection CNNs by Self Attention Distillation780
It is observed that attention maps extracted from a model trained to a reasonable level would encode rich contextual information that can be used as a form of ‘free’ supervision for further representation learning through performing top- down and layer-wise attention distillation within the net- work itself.
- 780
- 722
- Exploiting Diffusion Prior for Real-World Image Super-Resolution715
A novel approach to leverage prior knowledge encapsulated in pre-trained text-to-image diffusion models for blind super-resolution by employing the time-aware encoder can achieve promising restoration results without altering the pre-trained synthesis model, thereby preserving the generative prior and minimizing training cost.
- Non-local recurrent network for image restoration706
A non- local recurrent network (NLRN) is proposed as the first attempt to incorporate non-local operations into a recurrent neural network (RNN) for image restoration and achieves superior results to state-of-the-art methods with much fewer parameters.
- BasicVSR: The Search for Essential Components in Video Super-Resolution and Beyond698
A succinct pipeline is shown that achieves appealing improvements in terms of speed and restoration quality in comparison to many state-of-the-art algorithms and can serve as strong baselines for future VSR approaches.
- BasicVSR++: Improving Video Super-Resolution with Enhanced Propagation and Alignment687
This study redesigns BasicVsr by proposing second-order grid propagation and flow-guided deformable alignment, and shows that by empowering the re-current framework with enhanced propagation and align-ment, one can exploit spatiotemporal information across misaligned video frames more effectively.
- Region Proposal by Guided Anchoring685
This paper presents an alternative scheme, named Guided Anchoring, which leverages semantic features to guide the anchoring, and jointly predicts the locations where the center of objects of interest are likely to exist as well as the scales and aspect ratios at different locations.
- Feature Mining for Localised Crowd Counting677
This paper presents a single regression model based approach that is able to estimate people count in spatially localised regions and is more scalable without the need for training a large number of regressors proportional to the number of local regions.
- Semantic Image Segmentation via Deep Parsing Network673
This paper addresses semantic image segmentation by incorporating rich information into Markov Random Field (MRF), including high-order relations and mixture of label contexts by proposing a Convolutional Neural Network (CNN), namely Deep Parsing Network (DPN), which enables deterministic end-to-end computation in a single forward pass.
- 668
- Low-Light Image and Video Enhancement Using Deep Learning: A Survey655
This paper proposes a low-light image and video dataset, in which the images and videos are taken by different mobile phones’ cameras under diverse illumination conditions, and provides a unified online platform that covers many popular LLIE methods.
- DeeperForensics-1.0: A Large-Scale Dataset for Real-World Face Forgery Detection648
The on-going effort of constructing a large- scale benchmark for face forgery detection is presented, with 60, 000 videos constituted by a total of 17.6 million frames, 10 times larger than existing datasets of the same kind.
- From Facial Parts Responses to Face Detection: A Deep Learning Approach595
A novel deep convolutional network (DCN) that achieves outstanding performance on FDDB, PASCAL Face, and AFW, and it is shown that despite the use of DCN, the network can achieve practical runtime speed.
- 542
- MEAD: A Large-Scale Audio-Visual Dataset for Emotional Talking-Face Generation522
An emotional talking-face generation baseline that enables the manipulation of both emotion and its intensity and could be used for a number of research projects including conditional generation, cross-modal understanding and expression recognition.
- Focal Frequency Loss for Image Reconstruction and Synthesis520
It is shown that narrowing gaps in the frequency domain can ameliorate image reconstruction and synthesis quality further, and a novel focal frequency loss is proposed, which allows a model to adaptively focus on frequency components that are hard to synthesize by down-weighting the easy ones.
- 488
- Pedestrian Attribute Recognition At Far Distance479
A new pedestrian attribute dataset is released, which is by far the largest and most diverse of its kind and it is shown that the large-scale dataset facilitates the learning of robust attribute detectors with good generalization performance.
- Pose-Controllable Talking Face Generation by Implicitly Modularized Audio-Visual Representation474
This paper proposes a clean yet effective framework to generate pose-controllable talking faces whose poses are controllable by other videos and has multiple advanced capabilities including extreme view robustness and talking face frontalization.
- Exploiting Deep Generative Prior for Versatile Image Restoration and Manipulation458
This work presents an effective way to exploit the image prior captured by a generative adversarial network (GAN) trained on large-scale natural images and allows the generator to be fine-tuned on-the-fly in a progressive manner regularized by feature distance obtained by the discriminator in GAN.
- 454
- DeepID-Net: Deformable deep convolutional neural networks for object detection452
The proposed approach improves the mean averaged precision obtained by RCNN, which was the state-of-the-art, from 31% to 50.3% on the ILSVRC2014 detection test set.
- Self-Supervised Representation Learning: Introduction, advances, and challenges442
This article introduces this vibrant area, including key concepts, the four main families of approaches and associated state-of-the-art techniques, and how self-supervised methods are applied to diverse modalities of data.
- Learning Deep Representation for Face Alignment with Auxiliary Attributes435
A novel tasks-constrained deep model is formulated, which not only learns the inter-task correlation but also employs dynamic task coefficients to facilitate the optimization convergence when learning multiple complex tasks.
- Learning and Transferring Multi-task Deep Representation for Face Alignment.435
A novel tasks-constrained deep model is formulated, which not only learns the inter-task correlation but also employs dynamic task coefficients to facilitate the optimization convergence when learning multiple complex tasks.
- Depth Map Super-Resolution by Deep Multi-Scale Guidance376
A new method to address the problem of depth map super resolution in which a high-resolution (HR) depth map is inferred from a LR depth map and an additional HR intensity image of the same scene is presented.
- SparseNeRF: Distilling Depth Ranking for Few-shot Novel View Synthesis366
This work presents a new Sparse-view NeRF (SparseNeRF) framework that exploits depth priors from real-world inaccurate observations and outperforms all state-of-the-art few-shot NeRF methods on standard LLFF and DTU datasets.
- 360
- Deep Imbalanced Learning for Face Recognition and Attribute Prediction354
Cluster-based Large Margin Local Embedding (CLMLE), when combined with a simple k-nearest cluster algorithm, shows significant improvements in accuracy over existing methods on both face recognition and face attribute prediction tasks that exhibit imbalanced class distribution.
- Bailando: 3D Dance Generation by Actor-Critic GPT with Choreographic Memory351
A novel music-to-dance framework with a choreographic memory that learns to summarize meaningful dancing units from 3D pose sequence to a quantized codebook, and an actor-critic Generative Pre-trained Transformer (GPT) that composes these units to a fluent dance coherent to the music.
- Rerender A Video: Zero-Shot Text-Guided Video-to-Video Translation350
A novel zero-shot text-guided video-to-video translation framework to adapt image models to videos that achieves global style and local texture temporal consistency at a low cost (without re-training or optimization).
- 349
- Knowledge Distillation Meets Self-supervision348
A more general and model-agnostic approach for extracting "richer dark knowledge" from the pre-trained teacher model, and it is shown that the seemingly different self-supervision task can serve as a simple yet powerful solution for distillation.
- Transformer-Based Visual Segmentation: A Survey342
This survey provides a thorough overview of transformer-based visual segmentation, summarizing recent advancements and presents a meta-architecture that unifies all recent transformer-based approaches.
- Open-Vocabulary DETR with Conditional Matching336
This work proposes a novel open-vocabulary detector based on DETR -- hence the name OV-DETR, which, once trained, can detect any object given its class name or an exemplar image and achieves non-trivial improvements over current state of the arts.
- Reading Scene Text in Deep Convolutional Sequences322
A deep recurrent model is developed to robustly recognize the generated CNN sequences, departing from most existing approaches recognising each character independently, achieving impressive results on several benchmarks, advancing the state-of-the-art substantially.
- From Facial Expression Recognition to Interpersonal Relation Prediction317
This work designs an effective multitask network that is capable of learning from rich auxiliary attributes such as gender, age, and head pose, beyond just facial expression data and uses the expression recognition network as branches for a Siamese model to predict inter-personal relation.
- Person Re-identification: What Features Are Important?315
This study shows that certain features play more important role than others under different circumstances, and proposes a novel unsupervised approach for learning a bottom-up feature importance, so features extracted from different individuals are weighted adaptively driven by their unique and inherent appearance attributes.
- Seesaw Loss for Long-Tailed Instance Segmentation314
This work proposes Seesaw Loss to dynamically re-balance gradients of positive and negative samples for each category, with two complementary factors, i.e., mitigation factor and compensation factor, which obtains significant gains over Cross-Entropy Loss and achieves state-of-the-art performance on LVIS dataset without bells and whistles.
- MeViS: A Large-scale Benchmark for Video Segmentation with Motion Expressions305
The results show that current RVOS methods cannot effectively address motion expression-guided video segmentation, and a large-scale dataset called MeViS is proposed, which contains numerous motion expressions to indicate target objects in complex environments.
- GLEAN: Generative Latent Bank for Large-Factor Image Super-Resolution305
This work shows that pre-trained Generative Adversarial Networks (GANs), e.g., StyleGAN, can be used as a latent bank to improve the restoration quality of large-factor image super-resolution (SR) and shows clear improvements in terms of fidelity and texture faithfulness in comparison to existing methods.
- Audio-Driven Emotional Video Portraits300
EVP is presented, a system for synthesizing high-quality video portraits with vivid emotional dynamics driven by audios, using the Cross-Reconstructed Emotion Disentanglement technique to decompose speech into two decoupled spaces.
- Be Your Own Prada: Fashion Synthesis with Structural Coherence297
A novel and effective approach for generating new clothing on a wearer through generative adversarial learning, given an input image of a person and a sentence describing a different outfit, which “redresses” the person as desired, while keeping the wearer and her/his pose unchanged.
- 290
- Deep Flow-Guided Video Inpainting285
This work first synthesizes a spatially and temporally coherent optical flow field across video frames using a newly designed Deep Flow Completion network, then uses the synthesized flow fields to guide the propagation of pixels to fill up the missing regions in the video.
- Not All Pixels Are Equal: Difficulty-Aware Semantic Segmentation via Deep Layer Cascade281
A novel deep layer cascade (LC) method to improve the accuracy and speed of semantic segmentation and is an end-to-end trainable framework, allowing joint learning of all sub-models.
- PolyNet: A Pursuit of Structural Diversity in Very Deep Networks279
This work presents a new family of modules, namely the PolyInception, which can be flexibly inserted in isolation or in a composition as replacements of different parts of a network, and demonstrates substantial improvements over the state-of-the-art on the ILSVRC 2012 benchmark.
- LEDNet: Joint Low-Light Enhancement and Deblurring in the Dark258
This work introduces a novel data synthesis pipeline that models realistic low-light blurring degradations and presents the first large-scale dataset, LOL-Blur, and an effective network, named LEDNet, to perform joint low-light enhancement and deblurring.
- Point-to-Voxel Knowledge Distillation for LiDAR Semantic Segmentation256
This article addresses the problem of distilling knowledge from a large teacher model to a slim student network for LiDAR semantic segmentation by proposing the Point-to-Voxel Knowledge Distillation (PVD), which transfers the hidden knowledge from both point level and voxel level.
- 256
- Prime Sample Attention in Object Detection243
The notion of Prime Samples, those that play a key role in driving the detection performance are proposed, and a simple yet effective sampling and learning strategy called PrIme Sample Attention (PISA) is developed that directs the focus of the training process towards such samples.
- Deep Cascaded Bi-Network for Face Hallucination240
A novel framework for hallucinating faces of unconstrained poses and with very low resolution, which allows exceptional hallucination quality on in-the-wild low-res faces with significant pose and illumination variations is presented.
- Robust Multi-Modality Multi-Object Tracking236
This study designs a generic sensor-agnostic multi-modality MOT framework (mmMOT), where each modality is capable of performing its role independently to preserve reliability, and could further improving its accuracy through a novel multi- modality fusion module.
- 235
- Iterative Prompt Learning for Unsupervised Backlit Image Enhancement230
A novel unsupervised backlit image enhancement method by exploring the potential of Contrastive Language-Image Pre-Training (CLIP) for pixel-level image enhancement, which outperforms state-of-the-art methods in terms of visual quality and generalization ability.
- ReenactGAN: Learning to Reenact Faces via Boundary Transfer229
The proposed method, known as ReenactGAN, is capable of transferring facial movements and expressions from an arbitrary person's monocular video input to a target person’s video, and can perform photo-realistic face reenactment.
- Crowd Counting and Profiling: Methodology and Evaluation226
This study describes and compares the state-of-the-art methods for video imagery based crowd counting, and provides a systematic evaluation of different methods using the same protocol.
- Embedding Fourier for Ultra-High-Definition Low-Light Image Enhancement222
This work proposes a new solution, UHDFour, that embeds Fourier transform into a cascaded network and contributes the first real UHD LLIE dataset, \textbf{UHD-LL}, that contains 2,150 low-noise/normal-clear 4K image pairs with diverse darkness and noise levels captured in different scenarios.
- Online Deep Clustering for Unsupervised Representation Learning220
This work proposes Online Deep Clustering (ODC) that performs clustering and network update simultaneously rather than alternatingly, and designs and maintains two dynamic memory modules, i.e., samples memory to store samples' labels and features, andCentroids memory for centroids evolution.
- Lifelong Learning via Progressive Distillation and Retrospection219
A novel approach to lifelong learning is proposed, which tries to seek a better balance between preservation and adaptation via two techniques: Distillation and Retrospection.
- The Devil of Face Recognition Is in the Noise217
This work contributes cleaned subsets of popular face databases, i.e., MegaFace and MS-Celeb-1M datasets, and builds a new large-scale noise-controlled IMDb-Face dataset, and investigates ways to improve data cleanliness, including a comprehensive user study on the influence of data labeling strategies to annotation accuracy.
- Dense Intrinsic Appearance Flow for Human Pose Transfer214
A novel approach for the task of human pose transfer, which aims at synthesizing a new image of a person from an input image of that person and a target pose, is presented, which addresses the issues of limited correspondences identified between keypoints only and invisible pixels due to self-occlusion.
- Unified Vision and Language Prompt Learning213
A systematic study on two representative prompt tuning methods, namely text prompt tuning and visual prompt tuning, and proposes a simple approach called Unified Prompt Tuning (UPT), which essentially learns a tiny neural network to jointly optimize prompts across different modalities.
- A Lightweight Optical Flow CNN —Revisiting Data Fidelity and Regularization212
LiteFlowNet2 is built on the foundation laid by conventional methods and resembles the corresponding roles as data fidelity and regularization in variational methods and provides high flow estimation accuracy through early correction with seamless incorporation of descriptor matching.
- Crafting a Toolchain for Image Restoration by Deep Reinforcement Learning210
In comparison to conventional human-designed networks, RL-Restore is capable of restoring images corrupted with complex and unknown distortions in a more parameter-efficient manner using the dynamically formed toolchain1.
- Video Object Segmentation with Joint Re-identification and Attention-Aware Mask Propagation208
A deep recurrent network that is capable of segmenting and tracking objects in video simultaneously by their temporal continuity, yet able to re-identify them when they re-appear after a prolonged occlusion is formulated.
- Aligning Bag of Regions for Open-Vocabulary Object Detection200
This work proposes to align the embedding of bag of regions beyond individual regions as a bag, and surpasses the previous best results by 4.6 box AP50 and 2.8 mask AP on novel categories of open-vocabulary COCO and LVIS benchmarks, respectively.
- Deep Specialized Network for Illuminant Estimation199
This work proposes a novel Deep Specialized Network (DS-Net) that is adaptive to diverse local regions for estimating robust local illuminants and outperforms state-of-the-art methods such as [1, 2, 3].
- Understanding Deformable Alignment in Video Super-Resolution195
It is shown that deformable convolution can be decomposed into a combination of spatial warping and convolution, which reveals the commonality of deformable alignment and flow-based alignment in formulation, but with a key difference in their offset diversity.
- Self-Supervised Scene De-Occlusion194
This paper makes the first attempt to address the problem of scene de-occlusion through a novel and unified framework that recovers hidden scene structures without ordering and amodal annotations as supervisions, via Partial Completion Network (PCNet)-mask (M) and -content (C), that learn to recover fractions of object masks and contents, respectively, in a self-supervised manner.
- Investigating Tradeoffs in Real-World Video Super-Resolution185
A stochastic degradation scheme that reduces up to 40% of training time without sacrificing performance is proposed and it is suggested that employing longer sequences rather than larger batches during training allows more effective uses of temporal information, leading to more stable performance during inference.
- 184
- Local similarity-aware deep feature embedding183
This paper introduces a Position-Dependent Deep Metric (PDDM) unit, which is capable of learning a similarity metric adaptive to local feature structure that can be used to select genuinely hard samples in a local neighborhood to guide the deep embedding learning in an online and robust manner.
- Contextual Object Detection with Multimodal Large Language Models182
This work presents ContextDET, a unified multimodal model that is capable of end-to-end differentiable modeling of visual-language contexts, so as to locate, identify, and associate visual objects with language inputs for human-AI interaction.
- Human Attribute Recognition by Deep Hierarchical Contexts178
This work trains a Convolutional Neural Network to select the most attribute-descriptive human parts from all poselet detections, and combines them with the whole body as a pose-normalized deep representation, which surpasses competitive baselines on this dataset and other popular ones.
- Learning Social Relation Traits from Face Images178
A deep model is proposed that learns a rich face representation to capture gender, expression, head pose, and age-related attributes, and then performs pairwise-face reasoning for relation prediction, and forms a new network architecture with a bridging layer to leverage the inherent correspondences among these datasets.
- 177
- Upscale-A-Video: Temporal-Consistent Diffusion Model for Real-World Video Super-Resolution175
Upscale-A-Video is introduced, a text-guided latent diffusion framework for video upscaling that surpasses existing methods in both synthetic and real-world benchmarks, as well as in AI-generated videos, showcasing impressive visual realism and temporal consistency.
- Deep Learning Markov Random Field for Semantic Segmentation174
This paper addresses semantic segmentation by incorporating high-order relations and mixture of label contexts into MRF by proposing a Convolutional Neural Network (CNN), namely Deep Parsing Network (DPN), which enables deterministic end-to-end computation in a single forward pass.
- Everybody’s Talkin’: Let Me Talk as You Want172
A method to edit a target portrait footage by taking a sequence of audio as input to synthesize a photo-realistic video, which is end-to-end learnable and robust to voice variations in the source audio.
- DifFace: Blind Face Restoration With Diffused Error Contraction171
This work proposes a novel method named DifFace that is capable of coping with unseen and complex degradations more gracefully without complicated loss designs and can contract the error of the restoration backbone and thus makes this method more robust to unknown degradations.
- Talk-to-Edit: Fine-Grained Facial Editing via Dialog171
Talk-to-Edit is proposed, an interactive facial editing framework that performs fine-grained attribute manipulation through dialog between the user and the system that demonstrates the superiority of the framework in terms of the smoothness of fine- grained editing, the identity/attribute preservation, and the visual photorealism and dialog fluency.
- 171
- Pose-Robust Face Recognition via Deep Residual Equivariant Mapping169
A novel Deep Residual EquivAriant Mapping (DREAM) block is formulated, which is capable of adaptively adding residuals to the input deep representation to transform a profile face representation to a canonical pose that simplifies recognition.
- Side-Aware Boundary Localization for More Precise Object Detection167
This paper proposes an alternative approach, named as Side-Aware Boundary Localization (SABL), where each side of the bounding box is respectively localized with a dedicated network branch, to tackle the difficulty of precise localization in the presence of displacements with large variance.
- 165
- 161
- Zoom-Net: Mining Deep Feature Interactions for Visual Relationship Recognition160
This work presents two new pooling cells to encourage feature interactions, and sheds light on how one could resolve ambiguous and noisy object and predicate annotations by Intra-Hierarchical trees (IH-tree).
- Faceness-Net: Face Detection through Deep Facial Part Responses160
A deep convolutional neural network is proposed for face detection leveraging on facial attributes based supervision and achieves promising performance on popular benchmarks including FDDB, PASCAL Faces, AFW, and WIDER FACE.
- VideoBooth: Diffusion-based Video Generation with Image Prompts159
This paper proposes a feed-forward framework VideoBooth, a generalizable framework where a single model works for a wide range of image prompts with only feed-forward passes, and proposes to embed image prompts in a coarse-to-fine manner.
- CLIPSelf: Vision Transformer Distills Itself for Open-Vocabulary Dense Prediction159
An in-depth analysis of the region-language alignment in CLIP models is embarked on, which proposes an approach named CLIPSelf, which adapts the image-level recognition ability of CLIP ViT to local image regions without needing any region-text pairs.
- FASA: Feature Augmentation and Sampling Adaptation for Long-Tailed Instance Segmentation154
This work proposes a simple yet effective method, Feature Augmentation and Sampling Adaptation (FASA), that addresses the data scarcity issue by augmenting the feature space especially for rare classes by eliminating the need for inter-class transfer learning.
- Unsupervised 3D Shape Completion through GAN Inversion150
The proposed ShapeInversion uses a GAN pre-trained on complete shapes by searching for a latent code that gives a complete shape that best reconstructs the given partial input, and is capable of incorporating the rich prior captured in a well-trained generative model.
- Pastiche Master: Exemplar-Based High-Resolution Portrait Style Transfer149
A novel DualStyleGAN with flexible control of dual styles of the original face domain and the extended artistic portrait domain is introduced with superiority over state-of-the-art methods in high-quality portrait style transfer and flexible stylecontrol.
- Robust Reference-based Super-Resolution via C 2 -Matching149
The proposed C2-Matching significantly outperforms state of the arts by over 1dB on the standard CUFED5 benchmark and shows great generalizability on WR-SR dataset as well as robustness across large scale and rotation transformations.
- RGB-D Salient Object Detection with Cross-Modality Modulation and Selection149
A cross-modality feature modulation module to enhance feature representations by taking the depth features as prior, which models the complementary relations of RGB-D data and an adaptive feature selection module to select saliency-related features and suppress the inferior ones.
- Inter-Region Affinity Distillation for Road Marking Segmentation147
A novel knowledge distillation (KD) approach that can transfer ‘knowledge’ on scene structure more effectively from a teacher to a student model and consistently brings higher performance gains on all lightweight models, compared to previous distillation methods.
- TSIT: A Simple and Versatile Framework for Image-to-Image Translation145
A simple and versatile framework for image-to-image translation that allows multi-scale semantic structure information and style representation to be effectively captured and fused by the network, permitting the method to scale to various tasks in both unsupervised and supervised settings.
- Aesthetic-Driven Image Enhancement by Adversarial Learning145
The proposed EnhanceGAN framework is introduced, an adversarial learning based model that performs automatic image enhancement and is able to learn enhancement operators for the task of aesthetic-based image enhancement, and the effectiveness of a piecewise color enhancement module trained with weak supervision is shown.
- 142
- 142
- Learning to Cluster Faces on an Affinity Graph141
This work explores a novel approach, namely, learning to cluster instead of relying on hand-crafted criteria, and proposes a framework based on graph convolutional network, which combines a detection and a segmentation module to pinpoint face clusters.
- OMG-Seg: Is One Model Good Enough for all Segmentation?140
It is shown that OMG-Seg, a transformer-based encoder-decoder architecture with task-specific queries and outputs, can support over ten distinct segmentation tasks and yet significantly reduce computational and parameter overhead across various tasks and datasets.
- 139
- Text2Human139
Extensive quantitative and qualitative evaluations demonstrate that the proposed Text2Human framework can generate more diverse and realistic human images compared to state-of-the-art methods.
- EcoNAS: Finding Proxies for Economical Neural Architecture Search137
A reliable proxy is presented and a hierarchical proxy strategy is formulated that spends more computations on candidate networks that are potentially more accurate, while discards unpromising ones in early stage with a fast proxy to lead to an economical evolutionary-based NAS (EcoNAS), which achieves an impressive 400×search time reduction.
- A survey on heterogeneous face recognition: Sketch, infra-red, 3D and low-resolution133
This survey provides a comprehensive review of established techniques and recent developments in HFR, and offers a detailed account of datasets and benchmarks commonly used for evaluation.
- DNA-Rendering: A Diverse Neural Actor Repository for High-Fidelity Human-centric Rendering131
This work presents DNA-Rendering, a large-scale, high-fidelity repository of human performance data for neural actor rendering, and provides a large-scale and quantitative benchmark in full-scale, with multiple tasks to evaluate the existing progress of novel view synthesis, novel pose animation synthesis, and novel identity rendering methods.
- Efficient Diffusion Model for Image Restoration by Residual Shifting130
This study proposes a novel and efficient diffusion model for IR that significantly reduces the required number of diffusion steps, and establishes a Markov chain that facilitates the transitions between the high-quality and low-quality images by shifting their residuals, substantially improving the transition efficiency.
- CelebV-Text: A Large-Scale Facial Text-Video Dataset129
The effectiveness and potential of CelebV- Text are shown through extensive self-evaluation, and a benchmark is constructed with representative methods to standardize the evaluation of the facial text-to-video generation task.
- Deep Animation Video Interpolation in the Wild127
This work formally defines and study the animation video interpolation problem for the first time, and proposes an effective framework, AnimeInterp, with two dedicated modules in a coarse-to-fine manner, which outperforms existing state-of-the-art interpolation methods for animation videos.
- 126
- Video K-Net: A Simple, Strong, and Unified Baseline for Video Segmentation125
Video K-Net is presented, a simple, strong, and unified framework for fully end-to-end video panoptic seg-mentation that achieves state-of-the-art videoPanoptic segmentation results on Citscapes-VPS and KITTI-STEP without bells and whistles and can serve as a new flexible baseline in video segmentation.
- 125
- 122
- Masked Frequency Modeling for Self-Supervised Visual Pre-Training122
MFM demonstrates that, for both ViT and CNN, a simple non-Siamese framework can learn meaningful representations even using none of the following: (i) extra data, (ii) extra model, (iii) mask token.
- LN3Diff++: Scalable L atent N eural Fields D iffusion for Speedy 3D Generation117
A novel framework, LN3Diff++, designed to bridge the gap and facilitate fast, high-quality, and versatile conditional 3D generation and surpasses existing 3D diffusion methods in inference speed, requiring no per-instance optimization.
- Optimizing Video Object Detection via a Scale-Time Lattice117
This paper presents a unified framework that integrates detection, temporal propagation, and across-scale refinement on a Scale-Time Lattice and develops an adaptive scheme with the detector invoked on demand and thus obtain improved tradeoff.
- Path-Restore: Learning Network Path Selection for Image Restoration116
The results show that Path-Restore, a multi-path CNN with a pathfinder that can dynamically select an appropriate route for each image region is effective for real-world denoising, where the noise distribution varies across different regions on a single image.
- 116
- Open-Vocabulary SAM: Segment and Recognize Twenty-Thousand Classes Interactively114
The Open-Vocabulary SAM is introduced, a SAM-inspired model designed for simultaneous interactive segmentation and recognition, leveraging two unique knowledge transfer modules: SAM2CLIP and CLIP2SAM.
- Deep Network Interpolation for Continuous Imagery Effect Transition114
This work proposes a simple yet universal approach to attain a smooth control of diverse imagery effects in many low-level vision tasks, including image restoration, image-to-image translation, and style transfer through linear interpolation in the parameter space of two or more correlated networks.
- TransGaGa: Geometry-Aware Unsupervised Image-To-Image Translation114
A novel disentangle-and-translate framework to tackle the complex objects image-to-image translation task, which disentangles image space into a Cartesian product of the appearance and the geometry latent spaces and supports multimodal translation.
- The Re-identification Challenge110
This chapter presents an overview of recent progress and the state-of-the-art approaches to solving some of the fundamental challenges in person re-identification, benefiting from research in computer vision, pattern recognition and machine learning, and drawing insights from video analytics system design considerations for engineering practical solutions.
- 103
- 102
- Learning to Cluster Faces via Confidence and Connectivity Estimation100
This paper proposes a fully learnable clustering framework without requiring a large number of overlapped subgraphs, and transforms the clustering problem into two sub-problems, designed to estimate the confidence of vertices and the connectivity of edges, respectively.
- LiteFlowNet3: Resolving Correspondence Ambiguity for More Accurate Optical Flow Estimation98
LiteFlowNet3, a deep network consisting of two specialized modules, is introduced, to address the issue of outliers in the cost volume by amending each cost vector through an adaptive modulation prior to the flow decoding and improve the flow accuracy by exploring local flow consistency.
- Positional Encoding as Spatial Inductive Bias in GANs97
This work shows that SinGAN's impressive capability in learning internal patch distribution, to a large extent, is brought by the implicit positional encoding when using zero padding in the generators, and proposes a new multi-scale training strategy and demonstrates its effectiveness in the state-of-the-art unconditional generator StyleGAN2.
- Quantifying Facial Age by Posterior of Age Comparisons94
This work introduces a novel approach for annotating large quantity of in-the-wild facial images with high-quality posterior age distribution as labels and trains a network that jointly performs ordinal hyperplane classification and posterior distribution learning.
- 93
- Semi-Supervised Domain Generalization with Stochastic StyleMatch92
This work studies semi-supervised domain generalization (SSDG) and proposes StyleMatch, a simple approach that extends FixMatch with a couple of new ingredients tailored for SSDG: stochastic modeling for reducing overfitting in scarce labels, and multi-view consistency learning for enhancing domain generalization.
- CARAFE++: Unified Content-Aware ReAssembly of FEatures91
This work proposes unified Content-Aware ReAssembly of FEatures (CARAFE++), a universal, lightweight, and highly effective operator to fulfill this goal and shows great potential to serve as a strong building block for modern deep networks.
- On-the-fly feature importance mining for person re-identification86
This study proposes a novel unsupervised approach to bottom-up feature importance mining on-the-fly specific to each re-identification probe target image, so features extracted from different individuals are weighted adaptively driven by their salient and inherent appearance attributes.
- Consensus-Driven Propagation in Massive Unlabeled Data for Face Recognition83
It is shown that unlabeled face data can be as effective as the labeled ones, and Consensus-Driven Propagation (CDP) is proposed to tackle this challenging problem with two modules, the “committee” and the ”mediator”, which select positive face pairs robustly by carefully aggregating multi-view information.
- Harmonizing Visual Representations for Unified Multimodal Understanding and Generation82
Harmon is presented, a unified autoregres-sive framework that harmonizes understanding and generation tasks with a shared MAR encoder and achieves state-of-the-art image generation results on the GenEval, MJHQ30K and WISE benchmarks while matching the performance of methods with dedicated semantic encoders on image understanding benchmarks.
- 82
- Detecting and discriminating behavioural anomalies80
The proposed framework outperforms existing methods in detecting durational anomalies in complex behaviours and subtle anomalies that are difficult to detect when objects are viewed in isolation, allowing for a more effective detection of subtle anomalies in surveillance videos.
- Flare7K++: Mixing Synthetic and Real Datasets for Nighttime Flare Removal and Beyond79
Flare7K++ is introduced, the first comprehensive nighttime flare removal dataset, consisting of 962 real-captured flare images (Flare-R) and 7000 synthetic flares (Flare7K), and a new end-to-end pipeline to preserve the light source while removing lens flares.
- 79
- Playing for 3D Human Recovery78
This work contributes GTA-Human, a large-scale 3D human dataset generated with the GTA-V game engine, featuring a highly diverse set of subjects, actions, and scenarios, and discovers that synthetic data provides critical complements to the real data that is typically collected indoor.
- Text2Performer: Text-Driven Human Video Generation78
This work decomposes the VQVAE latent space into human appearance and pose representation in an unsupervised manner by utilizing the nature of human videos, and proposes continuous VQ-diffuser to sample a sequence of pose embeddings.
- SeedVR: Seeding Infinity in Diffusion Transformer Towards Generic Video Restoration75
SeedVR is a diffusion transformer designed to handle real-world video restoration with arbitrary length and resolution, and achieves highly-competitive performance on both synthetic and real-world benchmarks, as well as AI-generated videos.
- VToonify75
Extensive experimental results demonstrate the effectiveness of the proposed VToonify framework over existing methods in generating high-quality and temporally-coherent artistic portrait videos with flexible style controls.
- Panoptic Video Scene Graph Generation73
A high-quality PVSG dataset is contributed, which consists of 400 videos (289 third-person + 111 egocentric videos) with totally 150K frames labeled with panoptic segmentation masks as well as fine, temporal scene graphs.
- 72
- Security and Surveillance72
A significant application of video analysis and understanding is intelligent surveillance, which aims to interpret automatically human activity and detect unusual events that could pose a threat to public security and safety.
- EdgeSAM: Prompt-In-the-Loop Distillation for SAM71
The approach involves distilling the original ViT-based SAM image encoder into a purely CNN-based architecture, better suited for edge devices, and incorporates a lightweight module within the encoder, which achieves a 37-fold speed increase compared to the original SAM.
- Fresco: Spatial-Temporal Correspondence for Zero-Shot Video Translation71
This paper introduces FRESCO, intra-frame correspondence alongside inter-frame correspondence to establish a more robust spatial-temporal constraint, which ensures a more consistent transformation of semantically similar content across frames.
- Learning to Steer by Mimicking Features from Heterogeneous Auxiliary Networks71
This paper considerably improves the accuracy and robustness of predictions through heterogeneous auxiliary networks feature mimicking, a new and effective training method that provides us with much richer contextual signals apart from steering direction.
- Delving Deep Into Hybrid Annotations for 3D Human Recovery in the Wild71
This work focuses on the challenging task of in-the-wild 3D human recovery from single images when paired 3D annotations are not fully available, and shows that incorporating dense correspondence into in- the- wild 3Dhuman recovery is promising and competitive due to its high efficiency and relatively low annotating cost.
- Duolando: Follower GPT with Off-Policy Reinforcement Learning for Dance Accompaniment70
A GPT-based model is proposed, Duolando, which autoregressively predicts the subsequent tokenized motion conditioned on the coordinated information of the music, the leader's and the follower's movements, and devise an off-policy reinforcement learning strategy that allows the model to explore viable trajectories from out-of-distribution samplings, guided by human-defined rewards.
- MeViS: A Multi-Modal Dataset for Referring Motion Expression Video Segmentation69
This paper proposes a large-scale multi-modal dataset for referring motion expression video segmentation, focusing on segmenting and tracking target objects in videos based on language description of objects’ motions, and introduces MeViS, a dataset containing 33,072 human-annotated motion expressions in both text and audio.
- 69
- Temporally consistent video colorization with deep feature propagation and self-regularization learning67
A novel temporally consistent video colorization (TCVC) framework that effectively propagates frame-level deep features in a bidirectional way to enhance the temporal consistency of colorization and introduces a self-regularization learning (SRL) scheme to minimize the differences in predictions obtained using different time steps.
- TransEditor: Transformer-Based Dual-Space GAN for Highly Controllable Facial Editing67
This work proposes TransEditor, a novel Transformer-based framework to enhance interaction in a dual-space GAN for more controllable editing, and develops a new dual- space editing and inversion strategy to provide additional editing flexibility.
- Delving into Inter-Image Invariance for Unsupervised Visual Representations66
This work presents a comprehensive empirical study to better understand the role of inter-image invariance learning from three main constituting components: pseudo-label maintenance, sampling strategy, and decision boundary design, and introduces a unified and generic framework that supports the integration of unsupervised intra- and inter- image invarianceLearning.
- TransMoMo: Invariance-Driven Unsupervised Video Motion Retargeting66
A lightweight video motion retargeting approach TransMoMo that is capable of transferring motion of a person in a source video realistically to another video of a target person by exploiting invariance properties of three orthogonal factors of variation including motion, structure, and view-angle.
- 66
- PERF: Panoramic Neural Radiance Field From a Single Panorama65
PerF is presented, a 360-degree novel view synthesis framework that trains a panoramic neural radiance field from a single panorama and can be widely used for real-world applications, such as panorama-to-3D, text-to-3D, and 3D scene stylization applications.
- Removing Diffraction Image Artifacts in Under-Display Camera via Dynamic Skip Connection Network65
A physics-based image formation model is defined to better understand the degradation of Under-Display Camera images, and a new domain knowledge-enabled Dynamic Skip Connection Network (DISCNet) is designed to restore the UDC images.
- 64
- MosaicFusion: Diffusion Models as Data Augmenters for Large Vocabulary Instance Segmentation61
Experimental results on the challenging LVIS long-tailed and open-vocabulary benchmarks demonstrate that MosaicFusion can significantly improve the performance of existing instance segmentation models, especially for rare and novel categories.
- 61
- 59
- Control Color: Multimodal Diffusion-Based Interactive Image Colorization58
Control Color is introduced, a multi-modal colorization method that leverages the pre-trained Stable Diffusion model, offering promising capabilities in highly controllable interactive image colorization and addressing the long-standing issues of color overflow and inaccurate coloring.
- ReconfigISP: Reconfigurable Camera Image Processing Pipeline56
This study proposes a novel Reconfigurable ISP (ReconfigISP) whose architecture and parameters can be automatically tailored to specific data and tasks, and enables back-propagation for each module by training a differentiable proxy, allowing to leverage the popular differentiable neural architecture search and effectively search for the optimal ISP architecture.
- Task-Oriented Human-Object Interactions Generation with Implicit Neural Representations55
This work proposes TOHO: Task-Oriented Human-Object Interactions Generation with Implicit Neural Representations, which generates full human-object interaction motions to conduct specific tasks, given only the task type, the object, and a starting human status.
- Self-Supervised Learning via Conditional Motion Propagation55
This work designs a new learning-from-motion paradigm that solves the inherent motion ambiguity, and encourages the emergence of kinematically-sound representations that poss greater expressive power, and is successfully extended to several useful applications such as semi-automatic pixel-level annotation.
- Computation-Efficient Knowledge Distillation via Uncertainty-Aware Mixup54
It is shown that the UNcertainty-aware mIXup (UNIX) can serve as a clean yet effective solution and the redundancy of conventional knowledge distillation lies in the excessive learning of easy samples.
- Unsupervised Image-to-Image Translation with Generative Prior53
A novel framework to leverage the generative prior from pre-trained class-conditional GANs (e.g., BigGAN) to learn rich content correspondences across various domains, and proposes a novel coarse-to-fine scheme to capture a robust coarse-level content representation that can link objects at an abstract semantic level.
- Texture Memory-Augmented Deep Patch-Based Image Inpainting53
A new deep inpainting framework where texture generation is guided by a texture memory of patch samples extracted from unmasked regions is proposed, which has a novel design that allows texture memory retrieval to be trained end-to-end with the deep inPainting network.
- Development of fine-grained pill identification algorithm using deep convolutional network53
The superior performance of DCN underscores the potential of Deep Learning model in the application of pill identification and verification in the healthcare institutes.
- 52
- SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training51
This work proposes a one-step diffusion-based VR model, termed as SeedVR2, which performs adversarial VR training against real data, and introduces several enhancements to both model architecture and training procedures to handle the challenging high-resolution VR within a single step.
- OpenUni: A Simple Baseline for Unified Multimodal Understanding and Generation50
This report presents OpenUni, a simple, lightweight, and fully open-source baseline for unifying multimodal understanding and generation that bridging the off-the-shelf multimodal large language models (LLMs) and diffusion models through a set of learnable queries and a light-weight transformer-based connector.
- Nighttime Smartphone Reflective Flare Removal Using Optical Center Symmetry Prior50
This work proposes an optical center symmetry prior, which suggests that the reflective flare and light source are always symmetrical around the lens's optical center, and creates the first reflective flare removal dataset called BracketFlare, which contains diverse and realistic reflective flare patterns.
- SMPLest-X: Ultimate Scaling for Expressive Human Pose and Shape Estimation49
This work investigates the impact of scaling up EHPS towards a family of generalist foundation models, and finetuning strategy turns the generalist into specialist models, allowing them to achieve further performance boosts.
- 49
- Towards Language-Driven Video Inpainting via Multimodal Large Language Models48
This work introduces a new task - language-driven video inpainting, which uses natural language instructions to guide the inpainting process, integrating Multimodal Large Language Models to understand and execute complex language-based inpaintingrequests effectively.
- Joint Face Representation Adaptation and Clustering in Videos48
Experiments demonstrate that the proposed joint face representation adaptation and clustering approach generates character clusters with high purity compared to existing video face clustering methods, which are either based on deep face representation (without adaptation) or carefully engineered features.
- 46
- Stream-Based Active Unusual Event Detection46
Experimental results on busy public space videos show that with minimal human supervision, the approach outperforms existing supervised and unsupervised learning strategies in identifying unusual events.
- 46
- 45
- Crowd Saliency Detection via Global Similarity Structure45
A novel framework to identify and localize salient regions in a crowd scene, by transforming low-level features extracted from crowd motion field into a global similarity structure representation, which allows the discovery of the intrinsic manifold of the motion dynamics, which could not be captured by the low- level representation.
- 44
- GaussianAnything: Interactive Point Cloud Flow Matching For 3D Object Generation43
A novel 3D generation framework that addresses challenges with input formats, latent space design, and output representations, and incorporates a cascaded latent diffusion model for improved shape-texture disentanglement, offering scalable, high-quality 3D generation with an interactive Point Cloud-structured Latent space.
- 43
- NTIRE 2021 Challenge on Quality Enhancement of Compressed Video: Methods and Results43
This paper reviews the first NTIRE challenge on quality enhancement of compressed video, with a focus on the proposed methods and results, and gauge the state of the art of video quality enhancement.
- Learning Generative Structure Prior for Blind Text Image Super-resolution42
This work learns to encapsulate rich and diverse structures in a StyleGAN and exploit such generative structure priors for restoration and stores the discrete features for each character in a codebook to restrict the generative space of StyleGAN.
- Monocular 3D Reconstruction of Interacting Hands via Collision-Aware Factorized Refinements42
This paper makes the first attempt to reconstruct 3D interacting hands from monocular single RGB images via a two-stage framework that can generate 3D hand meshes with both precise 3D poses and minimal collisions.
- MVIP-NeRF: Multi-View 3D Inpainting on NeRF Scenes via Diffusion Prior41
This work proposes a novel approach called MVIP-NeRF that harnesses the potential of diffusion priors for NeRF inpainting, addressing both appearance and geometry aspects, and shows better appearance and geometry recovery than previous NeRF in painting methods.
- 41
- Position-Guided Point Cloud Panoptic Segmentation Transformer40
This work designs a robust Mixed-parameterized Positional Embedding to guide the segmentation process in LiDAR-based point cloud segmentation and outperforms previous state-of-the-art methods by 2.7% and 1.2% PQ on SemanticKITTI and nuScenes datasets, respectively.
- Betrayed by Captions: Joint Caption Grounding and Generation for Open Vocabulary Instance Segmentation40
This work devise a joint Caption Grounding and Generation (CGG) framework, which incorporates a novel grounding loss that only focuses on matching object nouns to improve learning efficiency and introduces a caption generation head that enables additional supervision and contextual modeling as a complementation to the grounding loss.
- 40
- Mix-and-Match Tuning for Self-Supervised Semantic Segmentation39
With the proposed M&M approach, for the first time, a self-supervision method can achieve comparable or even better performance compared to its ImageNet pretrained counterpart on both PASCAL VOC2012 dataset and CityScapes dataset.
- F-LMM: Grounding Frozen Large Multimodal Models38
F-LMM—grounding frozen off-the-shelf LMMs in human-AI conversations—a straightforward yet effective design based on the fact that word-pixel correspondences conducive to visual grounding inherently exist in the attention mechanism of well-trained LMMs.
- 38
- 38
- Tube-Link: A Flexible Cross Tube Framework for Universal Video Segmentation37
Tube-Link is a near-online approach that takes a short subclip as input and outputs the corresponding spatial-temporal tube masks, and introduces temporal contrastive learning to instance-wise discriminative features for tube-level association.
- Self-Supervised Geometry-Aware Encoder for Style-Based 3D GAN Inversion37
Extensive experiments show that the effective self-training scheme devised outperforms state-of-the-art inversion methods in both shape and texture reconstruction quality.
- Instance-Level Facial Attributes Transfer with Geometry-Aware Flow37
This work proposes the use of geometry-aware flow, which serves as a well-suited representation for modeling the transformation between instance-level facial attributes, and leverage the facial landmarks as the geometric guidance to learn the differentiable flows automatically, despite of the large pose gap existed.
- When StyleGAN Meets Stable Diffusion: a $\mathcal{W}_{+}$ Adapter for Personalized Image Generation36
The novel use of the extended StyleGAN embed-ding space $\mathcal{W}+$, to achieve enhanced identity preservation and disentanglement for diffusion models, and proposes new training objectives to balance the influences of both prompt and identity conditions.
- StyleGANEX: StyleGAN-Based Manipulation Beyond Cropped Aligned Faces36
This paper uses dilated convolutions to rescale the receptive fields of shallow layers in StyleGAN to enable real face inversion and manipulation, and introduces a corresponding encoder that provides the first-layer feature of the extended StyleGAN in addition to the latent style code.
- GLEAN: Generative Latent Bank for Image Super-Resolution and Beyond36
The method, Generative LatEnt bANk (GLEAN), goes beyond existing practices by directly leveraging rich and diverse priors encapsulated in a pre-trained GAN, and extends to different tasks including image colorization and blind image restoration.
- NTIRE 2021 Challenge on Video Super-Resolution36
This paper presents evaluation results from two competition tracks as well as the proposed solutions to the NTIRE 2021 Challenge on Video Super-Resolution, and develops conventional video SR methods focusing on the restoration quality.
- Modelling Multi-object Activity by Gaussian Processes36
Experimental results on a public traffic scene show that the GP models outperform DBNs in terms of anomaly sensitivity, noise robustness, and flexibility in modelling complex activity.
- 35
- 35
- 34
- Discriminative Sparse Neighbor Approximation for Imbalanced Learning34
This paper proposes a novel discriminative sparse neighbor approximation method that significantly outperforms the state-of-the-art methods that do not possess an imbalance handling mechanism, and is found to perform comparably or even better than recent deep learning methods by using hand-crafted features only.
- On the Generalization of BasicVSR++ to Video Deblurring and Denoising33
The proposed framework achieves compelling performance with great efficiency in various video restoration tasks including video deblurring and denoising and achieves comparable performance to Transformer-based approaches with up to 79% of parameter reduction and 44x speedup.
- Delving into High-Quality Synthetic Face Occlusion Segmentation Datasets33
This study proposes two occlusion generation techniques, Naturalistic Occlusion Generation (NatOcc), for producing high-quality naturalistic synthetic occluded faces; and Random Occlusions Generation (RandOcc), a more general synthetic Occluded data generation method.
- Network pruning via resource reallocation32
PEEL first constructs a predefined backbone and then conducts resource reallocation on it to shift parameters from less informative layers to more important layers in one round, thus amplifying the positive effect of these informative layers.
- MatAnyone: Stable Video Matting with Consistent Memory Propagation31
This work introduces a consistent memory propagation module via region-adaptive memory fusion, which adaptively integrates memory from the previous frame to ensure semantic stability in core regions while preserving fine-grained details along object boundaries.
- EdgeTAM: On-Device Track Anything Model31
This paper proposes EdgeTAM, which leverages a novel 2D Spatial Perceiver to reduce the computational cost and encodes the densely stored frame-level memories with a lightweight Transformer that contains a fixed set of learnable queries.
- Generating Aligned Pseudo-Supervision from Non-Aligned Data for Image Restoration in Under-Display Camera30
This work revisits the classic stereo setup for training data collection and mitigate the non-trivial domain discrepancy and spatial misalignment through a novel Transformer-based framework that generates well-aligned yet high-quality target data for the corresponding UDC input.
- Gaussian3Diff: 3D Gaussian Diffusion for 3D Full Head Synthesis and Editing30
The proposed approach leverages an implicit function representation of 3D human heads, employing 3D Gaussians anchored on a parametric face model, enabling effective utilization of the diffusion model for 3D head avatar generation.
- 29
- 27
- 27
- 27
- 26
- Learning Inclusion Matching for Animation Paint Bucket Colorization26
This work introduces a new learning-based inclusion matching pipeline, which directs the network to comprehend the inclusion relationships between segments rather than relying solely on direct visual correspondences, enabling more nuanced and accurate colorization.
- Scenimefy: Learning to Craft Anime Scene via Semi-Supervised Image-to-Image Translation26
This work proposes Scenimefy, a novel semi-supervised image-to-image translation framework that addresses the challenges of the complexity of the scenes, the unique features of anime style, and the lack of high-quality datasets to bridge the domain gap.
- MoCaNet: Motion Retargeting In-the-Wild via Canonicalization Networks26
A novel framework that brings the 3D motion retargeting task from controlled environments to in-the-wild scenarios and could serve as a disentangled and interpretable representation of human motion that benefits action analysis and motion retrieval.
- Learning from Multiple Sources for Video Summarisation26
This paper presents a novel unsupervised framework to learn jointly from both visual and independently-drawn non-visual data sources for discovering meaningful latent structure of surveillance video data, and investigates ways to cope with discrepant dimension and representation whilst associating these heterogeneous data sources.
- 26
- 26
- Mamba or RWKV: Exploring High-Quality and High-Efficiency Segment Anything Model25
This work designs a mixed backbone that contains convolution and RWKV operation, which achieves the best for both accuracy and efficiency and designs an efficient decoder to utilize the multiscale tokens to obtain high-quality masks.
- Towards Multi-Layered 3D Garments Animation25
This paper proposes a novel data-driven method, called LayersNet, to model garment-level animations as particle-wise interactions in a micro physics system to improve simulation efficiency by representing garments as patch-level particles in a two-level structural hierarchy.
- Reference-based Image and Video Super-Resolution via $C^{2}$-Matching25
A contrastive correspondence network, which learns transformation-robust correspondences using augmented views of the input image, and a dynamic aggregation module to address the potential misalignment issue between input images and reference images are designed.
- MessyTable: Instance Association in Multiple Camera Views25
An interesting and challenging dataset that features a large number of scenes with messy tables captured from multiple camera views, which challenges existing methods in mining subtle appearance differences, reasoning based on contexts, and fusing appearance with geometric cues for establishing an association.
- XOV-Action: Toward Generalizable Open-Vocabulary Action Recognition24
This work contributes a novel model named XOV-Action, which first proposes to capture diverse action-related concepts by learning diversified elaboration representations, which enables better generalization to open-set action categories and improves the generalization in unseen video domains.
- GP-UNIT: Generative Prior for Versatile Unsupervised Image-to-Image Translation24
The key idea of GP-UNIT is to distill the generative prior from pre-trained class-conditional GANs to build coarse-level cross-domain correspondences, and to apply the learned prior to adversarial translations to excavate fine-level correspondences.
- StyleInV: A Temporal Style Modulated Inversion Network for Unconditional Video Generation24
A novel motion generator design that uses a learning-based inversion network for GAN and enjoys the advantage of sparse training and naturally constrains the generation space of the motion generator with the inversion network guided by the initial frame, eliminating the need for heavy discriminators.
- 24
- Learning to Disambiguate by Asking Discriminative Questions24
This paper introduces the first large-scale dataset with over 10,000 carefully annotated images-question tuples to facilitate benchmarking and presents an effective method for visual discriminative question generation.
- CuDi: Curve Distillation for Efficient and Controllable Exposure Adjustment23
This work presents Curve Distillation, CuDi, for efficient and controllable exposure adjustment without the requirement of paired or unpaired data during training, which corrects both underexposed and overexposed photos with a single model.
- BeautyREC: Robust, Efficient, and Component-Specific Makeup Transfer23
The key insights of this study are modeling component-specific correspondence for local makeup transfer, capturing long-range dependencies for global makeupTransfer, and enabling efficient makeup transfer via a single-path structure.
- 23
- Discover and Learn New Objects from Documentaries22
A joint probabilistic framework is developed, where individual pieces of information, including video frames and subtitles, are brought together via both visual and linguistic links, and a weakly supervised learning algorithm is derived, where object model learning and training set mining are unified in an optimization procedure.
- Evaluating Feature Importance for Re-identification22
This study presents a comprehensive comparison and evaluation of up-to-date imagery features for person re-identification and introduces an unsupervised approach to learning a bottom-up measurement of feature importance.
- Activity understanding and unusual event detection in surveillance videos22
This thesis shows that learning time delayed pairwise activity correlations offers valuable contextual information for spatial and temporal topology inference of a camera network, robust person re-identification, and accurate activity-based video temporal segmentation.
- 21
- Pedestrian Color Naming via Convolutional Neural Network21
It is demonstrated that the Pedestrian Color Naming CNN (PCN-CNN) is superior over existing approaches in providing consistent color names on real-world pedestrian images, and the effectiveness of color descriptor extracted from PCN- CNN in complementing existing descriptors for the task of person re-identification is shown.
- 20
- Correlational Image Modeling for Self-Supervised Visual Pre-Training20
This work introduces Correlational Image Modeling (CIM), a novel and surprisingly effective approach to self-supervised visual pre-training that performs on par or better than the current state of the art on self- supervised and transfer benchmarks.
- Audio-Driven Dubbing for User Generated Contents via Style-Aware Semi-Parametric Synthesis20
This paper investigates an audio-driven dubbing method that is more feasible for User Generated Content (UGC) production, and develops a semi-parametric video renderer, which takes full advantage of the limited training data of the unseen speaker via a video-level retrieve-warp-refine pipeline.
- 20
- 20
- 4RC: 4D Reconstruction via Conditional Querying Anytime and Anywhere19
This work represents per-view 4D attributes in a minimally factorized form by decomposing them into base geometry and time-dependent relative motion, and introduces a novel encode-once, query-anywhere and anytime paradigm for 4D reconstruction from monocular videos.
- Thinking with Camera: A Unified Multimodal Model for Camera-Centric Understanding and Generation19
This work presents Puffin, a unified camera-centric multimodal model that extends spatial awareness along the camera dimension and introduces a novel paradigm that treats camera as language, enabling thinking with camera.
- MOWA: Multiple-in-One Image Warping Model19
This work proposes a Multiple-in-One image WArping model (named MOWA), the first work that solves multiple practical warping tasks in one single model and introduces a lightweight point-based classifier that predicts the task type and serves as prompts to modulate the feature maps for more accurate estimation.
- 19
- 19
- 19
- 3DEnhancer: Consistent Multi-View Diffusion for 3D Enhancement18
A novel 3D enhancement pipeline, dubbed 3DENHANCER, which employs a multi-view latent diffusion model to enhance coarse 3D inputs while preserving multi-view consistency and significantly outperforms existing methods.
- 18
- 18
- DeformToon3d: Deformable Neural Radiance Fields for 3D Toonification18
This paper proposes DeformToon3d, an effective toonification framework tailored for hierarchical 3D GAN that decomposes 3D toonification into subproblems of geometry and texture stylization to better preserve the original latent space.
- MIPI 2023 Challenge on Nighttime Flare Removal: Methods and Results18
The developed solutions in this challenge achieved state-of-the-art performance on Nighttime Flare Removal and a detailed description of all models developed is provided in this paper.
- Chasing the Tail in Monocular 3D Human Reconstruction With Prototype Memory18
This work proposes a prototype memory-augmented network, PM-Net, that effectively improves performances of predicting rare poses and significantly improves the models’ performances on rare poses while generating comparable results on other samples.
- Dense Siamese Network for Dense Unsupervised Learning18
DenseSiam benefits from the simple Siamese network and optimizes the consistency of different granularities, and proves that the simple location correspondence and interacted region embeddings are effective enough to learn the similarity.
- 18
- Merge or Not? Learning to Group Faces via Imitation Learning18
A novel face grouping framework that learns clustering strategy from ground-truth simulated behavior through imitation learning and makes sequential decision to dynamically decide when to merge two face instances/groups driven by short- and long-term rewards.
- 17
- Denoising as Adaptation: Noise-Space Domain Adaptation for Image Restoration17
This paper shows that it is possible to perform domain adaptation via the noise space using diffusion models and derives a meaningful diffusion loss that guides the restoration model in progressively aligning both restored synthetic and real-world outputs with a target clean distribution.
- 16
- Deep Geometrized Cartoon Line Inbetweening16
A new approach, AnimeInbet, is proposed, which geometrizes raster line drawings into graphs of endpoints and reframes the inbetweening task as a graph fusion problem with vertex repositioning, which can effectively capture the sparsity and unique structure of line drawings while preserving the details during inbetweening.
- 16
- Semi-Supervised and Long-Tailed Object Detection with CascadeMatch16
This paper proposes a novel pseudo-labeling-based detector called CascadeMatch, which surpasses existing state-of-the-art semi-supervised approaches—across a wide range of detection architectures—in handling long-tailed object detection.
- Siamese DETR16
Siamese DETR is presented, a Siamese self-supervised pretraining approach for the Transformer architecture in DETR that considers learning view-invariant and detection-oriented representations simultaneously through two complementary tasks, i.e., localization and discrimination, in a novel multi-view learning framework.
- MIPI 2024 Challenge on Few-shot RAW Image Denoising: Methods and Results15
The developed solutions in this challenge achieved state-of-the-art performance on Few-shot RAW Image Denoising and were summarized and reviewed in this paper.
- 15
- 15
- Correspondence Distillation from NeRF-Based GAN14
This paper exploits the rich semantics and structural priors encapsulated in a pre-trained NeRF-based GAN to build dense correspondences across different NeRFs of the same category, which can effectively enable many NeRF-based downstream applications such as texture transfer.
- From Local Temporal Correlation to Global Anomaly Detection14
The results demonstrate that the framework is able to detect not only anomalies that are visually obvious, but also those that are ambiguous or supported only by very weak visual evidence, e.g. those that can be easily missed by a human observer.
- 13
- 13
- 13
- Deep Learning for Scene-Independent Crowd Analysis13
This book chapter will introduce scene-independent crowd analysis with deep learning, and introduces multiple architectures of deep neural networks and training strategies to learn the feature representations for crowd analysis.
- 13
- Adaptive Window Pruning for Efficient Local Motion Deblurring12
This paper proposes a local motion deblurring vision Transformer built on adaptive window pruning Transformer blocks (AdaWPT), and removes local motion blur effectively without distorting sharp regions, demonstrated by its exceptional perceptual and quantitative improvements compared to state-of-the-art methods.
- 12
- 11
- DeepFakes Detection: the DeeperForensics Dataset and Challenge11
This chapter presents the on-going effort of constructing DeeperForensics-1.0, a large-scale forgery detection dataset, and describes the observations obtained from organizing DeeperForensics Challenge 2020, a real-world face forgery detection competition based on DeeperForensics-1.0.
- 11
- SA-LUT: Spatial Adaptive 4D Look-Up Table for Photorealistic Style Transfer10
Experiments demonstrate that SA-LUT substantially outperforms state-of-the-art methods, achieving a 66.7 % reduction in LPIPS score compared to 3D LUT approaches, while maintaining real-time performance at 16 FPS for video styling.
- 10
- 10
- 10
- 10
- MEAT: Multiview Diffusion Model for Human Generation on Megapixels with Mesh Attention9
This paper explores human multiview diffusion models at the megapixel level and introduces a solution called mesh attention to enable training at 10242 resolution, and presents valuable insights into applying multiview human motion videos for diffusion training, addressing the longstanding issue of data scarcity.
- MIPI 2024 Challenge on Demosaic for Hybridevs Camera: Methods and Results9
A total of 110 participants from both industrial and academic backgrounds contributed many valuable solutions to address the difficulty of the restoration of HybridEVS’s raw data, thus raising the reconstructed performance to a new height.
- 9
- Talking Faces: Audio-to-Video Face Generation9
This chapter discusses the definition and underlying challenges of the talking face generation, and presents an overview of recent progress, as well as introducing some widely used datasets and performance metrics.
- 9
- MIPI 2023 Challenge on RGBW Remosaic: Methods and Results8
This paper summarizes and reviews the RGBW Joint Remosaic and Denoise track on MIPI 2023, and describes the top three models developed in this challenge.
- On-Device Domain Generalization8
A systematic study of domain generalization (DG) for tiny neural networks finds that knowledge distillation (KD), a well-known technique for model compression, is much better for tackling the on-device DG problem than conventional DG methods.
- 8
- 8
- 8
- 8
- 8
- Surveillance video behaviour profiling and anomaly detection8
The results demonstrate that the framework is able to detect not only anomalies that are visually obvious, but also those that are ambiguous or supported only by very weak visual evidence, e.g. those that can be easily missed by a human observer.
- Point-In-Context: Understanding Point Cloud via In-Context Learning7
This paper introduces Point-In-Context (PIC), a pioneering framework for 3D point cloud understanding that leverages in-context learning with a standard transformer architecture that uniquely enables the execution of multiple tasks after a single, unified training phase, eliminating the need for fine-tuning.
- DoF-Gaussian: Controllable Depth-of-Field for 3D Gaussian Splatting7
DoF-Gaussian is introduced, a controllable depth-of-field method for 3D-GS that combines a lens-based imaging model based on geometric optics principles to control DoF effects and applies defocus-to-focus adaptation to minimize the gap in the circle of confusion.
- Light-SQ: Structure-aware Shape Abstraction with Superquadrics for Generated Meshes7
Light-SQ is presented, a novel superquadric-based optimization framework that explicitly emphasizes structure-awareness from three aspects, and enables efficient, high-fidelity, and editable shape abstraction with superquadrics for complex generated geometry, advancing the feasibility of 3D UGC creation.
- Trans-Adapter: A Plug-And-Play Framework for Transparent Image Inpainting7
This work proposes Trans-Adapter, a plug-and-play adapter that enables diffusion-based inpainting models to process transparent images directly and introduces LayerBench, a novel non-reference alpha edge quality evaluation metric for assessing transparency edge quality.
- LOC3DIFF: Local Diffusion for 3D Human Head Synthesis and Editing7
The proposed approach constructs an implicit representation of 3D human heads, anchored on a parametric face model, that represents semantic consistent head region by a local tri-plane, modulated by a 3D Gaussian.
- 7
- 7
- 7
- 6
- GausSim: Foreseeing Reality by Gaussian Simulator for Elastic Objects6
GausSim, a novel neural network-based simulator designed to capture the dynamic behaviors of realworld elastic objects represented through Gaussian kernels, is introduced, employing a hierarchical structure that further organizes kernels into CMSs with explicit formulations, enabling a coarse-to-fine simulation approach.
- MARS2 2025 Challenge on Multimodal Reasoning: Datasets, Methods, Results, Discussion, and Outlook6
The MARS2 2025 Challenge on Multi-modal Reasoning is reviewed, which aims to bring together different approaches in multimodal machine learning and LLMs via a large benchmark and better allows researchers to follow the state-of-the-art in this very dynamic area.
- Paint Bucket Colorization Using Anime Character Color Design Sheets6
This work introduces inclusion matching, which allows the network to understand the inclusion relationships between segments, rather than relying solely on direct visual correspondences, and significantly improves performance in both keyframe colorization and consecutive frame colorization.
- MIPI 2023 Challenge on RGB+ToF Depth Completion: Methods and Results6
The results of the competition are presented, the strengths and weaknesses of the top-performing methods are analyzed, and the implications of the findings for future research in RGB+sparse ToF depth completion are discussed.
- 6
- 6
- 6
- 6
- 6
- Dimensionality Reduction of Protein Mass Spectrometry Data Using Random Projection6
Investigation of the use of Random Projection in protein MS data dimensionality reduction finds that without sacrificing classification accuracy, performing RP prior to PCA significantly improves the computational time.
- Dimensionality Reduction of Protein Mass Spectrometry Data Using Random Projection6
Investigation of the use of Random Projection in protein MS data dimensionality reduction finds that without sacrificing classification accuracy, performing RP prior to PCA significantly improves the computational time.
- Direct 3D-Aware Object Insertion via Decomposed Visual Proxies5
DI (Decomposed Injection for Reference Composition and Target-integration), a novel framework that integrates interactive pose manipulation with high-fidelity 2D image synthesis to enable pose-controllable object insertion and outperforms previous methods in both geometric controllability and visual quality.
- 5
- Editorial: Special Issue on Deep Learning for Face Analysis5
This special issue of the International Journal of Computer Vision (IJCV) is a compilation of papers that present new deep learning approaches for various applications on face analysis that have gone through the rigorous process of multiple rounds of reviewing.
- Image and Video Understanding in Big Data5
It is clear that big visual data analysis and unerstanding impose significant scientific and technological chalenges since one not only should have to advance methods able to racefully scale to both big and diverse data whilst being computaionally cheap, but also should need to develop processes suitable or learning from big andiverse data without incurring prohibitive osts in human and monetary resources.
- Comparing Visual Feature Coding for Learning Disjoint Camera Dependencies5
The concept that visual features should be coded and selected automatically for robust and accurate time-delayed dependency learning is examined, and the Time Delayed Mutual Information (TDMI) is adopted due to its reported effectiveness and simplicity.
- Authenticating the identity of computer users with typing biometrics and the fuzzy min-max neural network5
The experimental outcomes positively demonstrate the potentials of using typing biome- trics and the FMM network to provide an additional security layer for the current password-based or PIN-based methods in authenticating the identity of computer users.
- Authenticating the Identity of Computer Users with Typing Biometrics and the Fuzzy Min-Max Neural Network( BIOMETRICS AND ITS APPLICATIONS)5
The experimental outcomes positively demonstrate the potentials of using typing biome- trics and the FMM network to provide an additional security layer for the current password-based or PIN-based methods in authenticating the identity of computer users.
- Enhanced Generative Structure Prior for Chinese Text Image Super-Resolution4
This paper introduces a high-quality text image SR framework designed to restore the precise strokes of low-resolution (LR) Chinese characters, and proposes a novel structure prior that offers structure-level guidance to enhance visual quality.
- Half-Physics: Enabling Kinematic 3D Human Model with Physical Interactions4
This work proposes a "half-physics"mechanism that transforms 3D kinematic motion into a physics simulation and maintains kinematic control over inherent SMPL-X poses while ensuring physically plausible interactions with scenes and objects, effectively eliminating penetration and unrealistic object dynamics.
- MIPI 2024 Challenge on Nighttime Flare Removal: Methods and Results4
The developed solutions in this challenge achieved state-of-the-art performance on Nighttime Flare Removal and the developed solutions in this challenge achieved state-of-the-art performance on Nighttime Flare Removal.
- 4
- BeautyREC: Robust, Efficient, and Content-preserving Makeup Transfer4
In what follows, this work details the network structure and parameters, including contentEncoder, style encoder, componentspecific correspondence, long-range dependencies, and image reconstruction, including how the component-specific transferred features and the global transferred features are concatenated with the corresponding decoder features.
- 4
- 4
- Feature Matters: A Stage-by-Stage Approach for Knowledge Transfer.4
It is argued that feature is the most important knowledge from teacher, and sufficient for student to achieve appealing performance by just learning similar features as teacher without any processing, and presented an efficient learning strategy, which is to make student mimic features of teacher stage by stage.
- Deep Learning Face Attributes for Detection and Alignment4
This chapter aims to provide an in-depth presentation of recent progress and the current state-of-the-art approaches to solving some of the fundamental challenges in face attribute recognition, particularly from the angle of deep learning.
- Generative Photographic Control for Scene-Consistent Video Cinematic Editing3
This paper proposes CineCtrl, the first video cinematic editing framework that provides fine control over professional camera parameters (e.g., bokeh, shutter speed), and introduces a decoupled cross-attention mechanism to disentangle camera motion from photographic inputs, allowing fine-grained, independent control without compromising scene consistency.
- 3
- MIPI 2023 Challenge on RGBW Fusion: Methods and Results3
This paper summarizes and reviews the RGBW Joint Fusion and Denoise track on MIPI 2023 and describes the models developed in this challenge, including PSNR, SSIM, LPIPS, and KLD.
- 3
- 3
- Use of Circle-Segments as a Data Visualization Technique for Feature Selection in Pattern Classification3
In this paper, the circle-segments method is proposed as a feature selection method to identify important input features before the entire data set is provided for learning with machine learning systems.
- AITTI: Learning Adaptive Inclusive Token for Text-to-Image Generation2
The core of the method is a lightweight adaptive mapping network, which can customize the inclusive tokens for the concepts to be de-biased, making the tokens generalizable to unseen concepts regardless of their original bias distributions.
- 2
- 2
- Interpret Vision Transformers as ConvNets with Dynamic Convolutions2
This paper interprets vision Transformers as ConvNets with dynamic convolutions, which enables us to characterize existing Transformers and dynamic ConvNet in a unified framework and compare their design choices side by side.
- 2
- 2
- 2
- 2
- AnyBokeh: Physics-Guided Any-to-Any Bokeh Editing with Optical Fingerprint Transfer1
Experiments on real-world benchmarks show that AnyBokeh achieves faithful and controllable editing across any-to-any bokeh editing, all-in-focus-to-bokeh rendering, and defocus deblurring, while avoiding all-in-focus reconstruction and test-time bokeh-level calibration commonly required by existing approaches.
- Digital Life Project 2: Open-source Autonomous 3D Characters on the Web1
DLP2 demonstrates how LLMs can be embodied in responsive 3D characters, offering a practical blueprint for real-time, emotionally adaptive digital interactions on the web.
- $$E^{3}DGE$$: Self-Supervised Geometry-Aware Encoder for Style-Based 3D GAN Inversion1
This paper addresses the challenge of 3D GAN inversion, focusing on predicting a latent code from a single 2D image to faithfully recover 3D shapes and textures, and introduces an efficient self-training scheme that does not rely on real-world 2D-3D pairs but instead utilizes proxy samples generated from a 3D GAN.
- 1
- 1
- 1
- 1
- 1
- –
- –
- LA-Sign: Looped Transformers with Geometry-aware Alignment for Skeleton-based Sign Language Recognition–
LA-Sign is proposed, a looped transformer framework with geometry-aware alignment for ISLR that achieves state-of-the-art results while using fewer unique layers, highlighting the effectiveness of recurrent latent refinement and geometry-aware representation learning for sign language recognition.
- Generate Your Talking Avatar from Video Reference–
Talking Avatar generation from Video Reference (TAVR), a novel framework that shifts the paradigm by leveraging cross-scene video inputs and benefits from flexible inference-time video referencing and consistently surpasses existing baselines both quantitatively and qualitatively is introduced.
- –
- SignDPO: Multi-level Direct Preference Optimisation for Skeleton-based Gloss-free Sign Language Translation–
SignDPO consistently outperforms state-of-the-art gloss-free methods and even rivals established gloss-based ones, and suggests that multi-level preference alignment is a powerful paradigm for bridging the gap between high-entropy skeletal trajectories and discrete linguistic semantics.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Multi-Source Video Summarisation.–
This paper presents a novel unsupervised framework to learn jointly from both visual and independently-drawn non-visual data sources for discovering meaningful latent structure of surveillance video data, and investigates ways to cope with discrepant dimension and representation whist associating these heterogeneous data sources.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.