Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Larry Steven DavisSuggested from co-authorship
- L.S. DavisSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works184 from public data
- Learning Deconvolution Network for Semantic Segmentation4,393
A novel semantic segmentation algorithm by learning a deep deconvolution network on top of the convolutional layers adopted from VGG 16-layer net, which demonstrates outstanding performance in PASCAL VOC 2012 dataset.
- Learning Multi-domain Convolutional Neural Networks for Visual Tracking2,726
A novel visual tracking algorithm based on the representations from a discriminatively trained Convolutional Neural Network using a large set of videos with tracking ground-truths to obtain a generic target representation.
- Large-Scale Image Retrieval with Attentive Deep Local Features867
An attentive local feature descriptor suitable for large-scale image retrieval, referred to as DELE (DEep Local Feature), based on convolutional neural networks, which are trained only with image-level annotations on a landmark image dataset.
- The Visual Object Tracking VOT2016 Challenge Results776
The Visual Object Tracking challenge VOT2016 goes beyond its predecessors by introducing a new semi-automatic ground truth bounding box annotation methodology and extending the evaluation system with the no-reset experiment.
- The Visual Object Tracking VOT2014 Challenge Results588
The Visual Object Tracking challenge 2014, VOT2014, aims at comparing short-term single-object visual trackers that do not apply pre-learned models of object appearance and introduces a new dataset with full annotation of targets by rotated bounding boxes and per-frame attribute.
- 491
- Domain-Specific Batch Normalization for Unsupervised Domain Adaptation445
A novel unsupervised domain adaptation framework based on domain-specific batch normalization in deep neural networks that achieves the state-of-the-art accuracy in the standard setting and the multi-source domain adaption scenario.
- Weakly Supervised Action Localization by Sparse Temporal Pooling Network401
This work proposes a weakly supervised temporal action localization algorithm on untrimmed videos using convolutional neural networks that attains state-of-the-art results on the THUMOS14 dataset and outstanding performance on ActivityNet1.3 even with its weak supervision.
- Channel Attention Is All You Need for Video Frame Interpolation381
A simple but effective deep neural network for video frame interpolation, which is end-to-end trainable and is free from a motion estimation network component, and achieves outstanding performance compared to the existing models with a component for optical flow computation.
- Local-Global Video-Text Interactions for Temporal Grounding366
This paper addresses the problem of text-to-video temporal grounding using a novel regression-based model that learns to extract a collection of mid-level features for semantic phrases in a text query, which corresponds to important semantic entities described in the query.
- Image Question Answering Using Convolutional Neural Network with Dynamic Parameter Prediction345
The proposed network-joint network with the CNN for ImageQA and the parameter prediction network-is trained end-to-end through back-propagation, where its weights are initialized using a pre-trained CNN and GRU.
- URVOS: Unified Referring Video Object Segmentation Network with a Large-Scale Benchmark328
The URVOS algorithm addresses the challenging problem by performing language-based object segmentation and mask propagation jointly using a single deep neural network with a proper combination of two attention models.
- Real-Time MDNet299
This work presents a fast and accurate visual tracking algorithm based on the multi-domain convolutional neural network (MDNet) that accelerates feature extraction procedure and learns more discriminative models for instance classification; it enhances representation quality of target and background by maintaining a high resolution feature map with a large receptive field per activation.
- InfoNeRF: Ray Entropy Minimization for Few-Shot Neural Volume Rendering279
An information-theoretic regularization technique for few-shot novel view synthesis based on neural im-plicit representation that minimizes potential reconstruction inconsistency that happens due to in-sufficient viewpoints by imposing the entropy constraint of the density in each ray.
- 278
- Learning to Optimize Domain Specific Normalization for Domain Generalization270
The state-of-the-art accuracy of the algorithm in the standard domain generalization benchmarks is demonstrated, as well as viability to further tasks such as multi-source domain adaptation anddomain generalization in the presence of label noise.
- Class-Incremental Learning by Knowledge Distillation with Adaptive Feature Consolidation267
This algorithm is based on knowledge distillation and provides a principled way to maintain the representations of old models while adjusting to new tasks effectively, which effectively alleviates the notorious catastrophic forgetting problem despite the limited accessibility of data in the previous tasks.
- 247
- 192
- 189
- Learning Transferrable Knowledge for Semantic Segmentation with Deep Convolutional Neural Network177
A novel weakly-supervised semantic segmentation algorithm based on Deep Convolutional Neural Network based on DCNN that exploits auxiliary segmentation annotations available for different categories to guide segmentations on images with only image-level class labels.
- Streamlined Dense Video Captioning171
A novel dense video captioning framework is proposed, which models temporal dependency across events in a video explicitly and leverages visual and linguistic context from prior events for coherent storytelling.
- 164
- Weakly Supervised Semantic Segmentation Using Web-Crawled Videos161
A novel algorithm for weakly supervised semantic segmentation based on image-level class labels only that substantially outperforms existing methods based on the same level of supervision and is even as competitive as the approaches relying on extra annotations.
- 159
- 156
- 155
- Weakly Supervised Semantic Segmentation Using Superpixel Pooling Network154
Superpixel Pooling Network (SPN) is proposed, which utilizes superpixel segmentation of input image as a pooling layout to reflect low-level image structure for learning and inferring semantic segmentation.
- 152
- Variable-Rate Deep Image Compression through Spatially-Adaptive Feature Transform143
The proposed approach successfully improves performance on image classification and text region quality preservation via task-aware quality map estimation without additional model training and achieves outstanding rate-distortion trade-off compared to the approaches based on multiple models that are optimized separately for several different target rates.
- 126
- 122
- CPlaNet: Enhancing Image Geolocalization by Combinatorial Partitioning of Maps119
A simple but effective algorithm, combinatorial partitioning, which generates a large number of fine-grained output classes by intersecting multiple coarse- grained partitionings of the earth, which achieves the state-of-the-art performance in location recognition on multiple benchmark datasets.
- 119
- 116
- MarioQA: Answering Questions by Watching Gameplay Videos115
It is shown that properly constructed datasets with events in various complexity levels are critical to learn effective models and improve overall performance and this work automatically generates a customized synthetic VideoQA dataset using Super Mario Bros. gameplay videos.
- On the Stability-Plasticity Dilemma of Class-Incremental Learning98
Surprisingly, the majority of class-incremental learning algorithms heavily favor stability over plasticity, to the extent that the feature extractor of a model trained on the initial set of classes is no less effective than that of the final incremental model.
- Text-Guided Attention Model for Image Captioning96
A text-guided attention model for image captioning, which learns to drive visual attention using associated captions using exemplar-based learning approach, which enables to describe a detailed state of scenes by distinguishing small or confusable objects effectively.
- Multi-Level Branched Regularization for Federated Learning83
A novel architectural regularization technique is proposed that constructs multiple auxiliary branches in each local model by grafting local and global subnetworks at several different levels and that learns the representations of the main pathway in the local model congruent to the auxiliary hybrid pathways via online knowledge distillation.
- 83
- 82
- Learning for Single-Shot Confidence Calibration in Deep Neural Networks Through Stochastic Inferences81
A novel variance-weighted confidence-integrated loss function that is composed of two cross-entropy loss terms with respect to ground-truth and uniform distribution, which are balanced by variance of stochastic prediction scores is designed.
- Attentive Semantic Alignment with Offset-Aware Correlation Kernels80
This paper introduces an attentive semantic alignment method that focuses on reliable correlations, filtering out distractors, and proposes an offset-aware correlation kernel that learns to capture translation-invariant local transformations in computing correlation values over spatial locations.
- Task-Aware Quantization Network for JPEG Image Compression77
A deep neural network is proposed to learn for JPEG image compression, which predicts image-specific optimized quantization tables fully compatible with the standard JPEG encoder and decoder and provides the capability to learn task-specific quantization tables in a principled way by adjusting the objective function of the network.
- Online Multi-target Tracking by Large Margin Structured Learning77
This work presents an online data association algorithm for multi-object tracking using structured prediction and provides a method to handle entering/leaving objects, short-term occlusions, and misdetections by introducing virtual agents--additional nodes in a bipartite graph.
- 76
- Class-Incremental Learning for Action Recognition in Videos72
This framework addresses catastrophic forgetting problem in the context of class-incremental learning for video recognition by introducing time-channel importance maps and exploiting the importance maps for learning the representations of incoming examples via knowledge distillation.
- 72
- 72
- Unsupervised Learning of Debiased Representations with Pseudo-Attributes66
This work first identifies pseudo-attributes based on the results from clustering performed in the feature embedding space even without an explicit bias attribute supervision, and employs a novel cluster-wise reweighting scheme to learn debiased representation.
- The Thermal Infrared Visual Object Tracking VOT-TIR2016 Challenge Results66
A significant general improvement of results has been observed, which partly compensate for the more difficult sequences introduced into the dataset in VOT-TIR2016.
- Traffic Accident Benchmark for Causality Recognition65
A brand new benchmark for analyzing causality in traffic accident videos by decomposing an accident into a pair of events, cause and effect, which mitigates the ambiguity of legal liabilities among agents engaged in the accident.
- 63
- Relaxed Contrastive Learning for Federated Learning61
A relaxed contrastive learning loss is introduced that imposes a divergence penalty on excessively similar sample pairs within each class, which prevents collapsed representations and enhances feature transferability, facilitating collaborative training and leading to significant performance improvements.
- Communication-Efficient Federated Learning with Accelerated Client Gradient58
This work proposes a simple but effective federated learning framework, which improves the consistency across clients and facilitates the convergence of the server model by making the server broadcast a global model with a lookahead gradient.
- Exemplar-Based Open-Set Panoptic Segmentation Network58
A novel exemplar-based open-set panoptic segmentation network (EOPSN) inspired by exemplar theory is proposed, which identifies a new class based on exemplars, which are identified by clustering and employed as pseudoground-truths.
- 57
- 54
- Modeling and segmentation of floating foreground and background in videos54
An on-line algorithm to segment foreground from background in videos captured by a moving camera using temporal model propagation and spatial model composition is proposed, and likelihood maps are computed based on the models.
- 52
- Towards Oracle Knowledge Distillation with Neural Architecture Search50
It is shown that searching for a new student model is effective in both accuracy and memory size and that the searched models often outperform their teacher models thanks to neural architecture search with oracle knowledge distillation.
- Generalized Background Subtraction Using Superpixels with Label Integrated Motion Estimation50
An online background subtraction algorithm with superpixel-based density estimation for videos captured by moving camera and label integrated motion estimation that makes propagated appearance models more accurate compared to existing methods since the models are constructed on visually coherent regions.
- Extracting Moving People from Internet Videos50
This work proposes a fully automatic framework to detect and extract arbitrary human motion volumes from real-world videos collected from YouTube and applies a top-down pictorial structure model to complete the extraction of the humans in arbitrary motion.
- Continual Learning by Asymmetric Loss Approximation With Single-Side Overestimation49
This work proposes a novel approach to continual learning by approximating a true loss function using an asymmetric quadratic function with one of its sides overestimated, which achieves the state-of-the-art accuracy that is even close to the upper-bound performance on several challenging benchmark datasets.
- 49
- 46
- Object Localization based on Structural SVM using Privileged Information46
This work tackles object localization problem based on a novel structural SVM using privileged information, where an alternating loss-augmented inference procedure is employed to handle the term in the objective function corresponding to privileged information.
- 45
- 45
- 43
- Bayesian Filtering and Integral Image for Visual Tracking43
The use of kernel-based Bayesian filtering for the tracking control procedure, and feature-based tracking to improve the observation process of tracking are described.
- Hierarchical Attention Networks.42
A novel attention network, which accurately attends to target objects of various scales and shapes in images through multiple stages, which outperforms traditional attention methods in various attribute prediction tasks.
- Learning to Specialize with Knowledge Distillation for Visual Question Answering40
This work presents a principled algorithm to learn specialized models with knowledge distillation under a multiple choice learning (MCL) framework, where training examples are assigned dynamically to a subset of models for updating network parameters.
- 40
- 40
- 40
- Online Graph-Based Tracking38
This work proposes a visual tracking algorithm based on more general graphical models, where multiple previous frames contribute to computing the posterior in the current frame and edges between frames are created upon inter-frame trackability.
- 36
- The Tenth NTIRE 2025 Image Denoising Challenge Report34
An overview of the NTIRE 2025 Image Denoising Challenge is presented, highlighting the proposed methodologies and corresponding results, providing insights into the current state-of-the-art in image denoising.
- 34
- Real-Time Object Tracking via Meta-Learning: Efficient Model Adaptation and One-Shot Channel Pruning33
A novel meta-learning framework for real-time object tracking with efficient model adaptation and channel pruning that greatly improves tracking performance by accelerating the convergence of online learning and reducing the cost of feature computation.
- Context-Aware Zero-Shot Recognition32
This work aims to understand the identity of the novel objects in an image surrounded by the known objects using the inter-object relation prior, and leverage the visual context and the geometric relationships between all pairs of objects in a single image, and capture the information useful to infer unseen categories.
- 31
- Information-Theoretic Bias Reduction via Causal View of Spurious Correlation30
A novel debiasing framework against the algorithmic bias is designed, which incorporates a bias regularization loss derived by the proposed information-theoretic bias measurement approach and is validated in diverse realistic scenarios.
- 30
- 28
- Visual Tracking by Sampling Tree-Structured Graphical Models28
This work proposes a novel offline tracking algorithm by identifying a tree-structured graphical model, where it is formulated a unified framework to optimize tree structure and track a target in a principled way, based on MCMC sampling.
- 28
- 27
- 26
- Learning Semantic Segmentation from Multiple Datasets with Label Shifts25
This paper proposes UniSeg, an effective approach to automatically train models across multiple datasets with differing label spaces, without any manual relabeling efforts, and proposes two losses that account for conflicting and co-occurring labels to achieve better generalization performance in unseen domains.
- Reinforcing an Image Caption Generator Using Off-Line Human Feedback25
The empirical evidence indicates that the proposed policy gradient method learns to generalize the human raters' judgments to a previously unseen set of images, as judged by a different set of human judges, and additionally on a different, multi-dimensional side-by-side human evaluation procedure.
- Interactive motion effects design for a moving object in 4D films24
An algorithm that allows for the rapid design of motion effects for 4D films based on a viewer-centered rendering strategy that matches chair motion to the movement of a viewer's visual attention is presented.
- Image Retrieval with Deep Local Features and Attention-based Keypoints.24
This work introduces a local feature descriptor for large-scale image retrieval applications, called DELF (DEep Local Feature), which is based on convolutional neural networks, which are trained without objectand patch-level annotations on a landmark image dataset.
- 24
- Real-Time Subspace-Based Background Modeling Using Multi-channel Data24
This work proposes an efficient background modeling and subtraction algorithm using 2-Dimensional Principal Component Analysis (2DPCA), where multi-channel data are naturally integrated in eigenbackground framework with no additional dimensionality.
- NTIRE 2025 Challenge on Image Super-Resolution ($\times 4$): Methods and Results23
The NTIRE 2025 image super-resolution ($\times 4$) challenge, one of the associated competitions of the 10th NTIRE Workshop at CVPR 2025, aims to recover high-resolution images from low-resolution counterparts generated through bicubic downsampling with a $\times 4$ scaling factor.
- 22
- Probabilistic fusion-based parameter estimation for visual tracking22
A robust fusion algorithm to estimate the complex motion parameters using variable-bandwidth mean-shift with correlation-based uncertainty modeling and fusion of individual components is proposed and the motion parameter that is robust to outliers can be detected with variable- bandwidth density-based fusion (VBDF) algorithm.
- AIM 2019 Challenge on Video Temporal Super-Resolution: Methods and Results20
This paper reviews the first AIM challenge on video temporal super-resolution (frame interpolation) with a focus on the proposed solutions and results, and employs the REDS_VTSR dataset derived from diverse videos captured in a hand-held camera for training and evaluation purposes.
- Combinatorial Inference against Label Noise19
This work proposes a unique classification framework of constructing multiple models in heterogeneous coarse-grained meta-class spaces and making joint inference of the trained models for the final predictions in the original (base) class space.
- 19
- Pooling Revisited: Your Receptive Field is Suboptimal18
DynOPool determines the proper resolution of a feature map by learning the desirable size and shape of its receptive field, which allows an operator in a deeper layer to observe an input image in the optimal scale.
- Revisiting Machine Unlearning with Dimensional Alignment17
A novel evaluation metric for machine unlearning is introduced, coined dimensional alignment, which measures the alignment between the eigenspaces of the forget and retain sets, and is incorporated as a regularizer loss to develop a robust and stable unlearning framework.
- Transfer Learning via Unsupervised Task Discovery for Visual Question Answering17
This work employs linguistic knowledge sources such as structured lexical database and visual descriptions for unsupervised task discovery, and transfer a learned task conditional visual classifier as an answering unit in a visual question answering model.
- Learning to Select Pre-Trained Deep Representations with Bayesian Evidence Framework17
The proposed Bayesian evidence framework provides a good solution to identify the best ensemble of heterogeneous CNNs through a greedy algorithm and illustrates the state-of-the-art performance consistently in terms of prediction accuracy and modeling efficiency.
- Unsupervised Co-Activity Detection from Multiple Videos Using Absorbing Markov Chain17
A simple but effective unsupervised learning algorithm to detect a common activity (co-activity) from a set of videos, which is formulated using absorbing Markov chain in a principled way and illustrates outstanding performance quantitatively and qualitatively.
- 17
- 16
- NICE: CVPR 2023 Challenge on Zero-shot Image Captioning14
Through the challenge, the image captioning models were tested using a new evaluation dataset that includes a large variety of visual concepts from many domains that will contribute to the improvement of AI models on various vision-language tasks.
- 14
- 14
- Online Video Segmentation by Bayesian Split-Merge Clustering14
An online video segmentation algorithm based on a novel nonparametric Bayesian clustering method called Bayesian Split-Merge Clustering (BSMC), which can efficiently cluster dynamically changing data through split and merge processes at each time step.
- Seam carving with forward gradient difference maps14
The proposed technique preserves straight lines and regular shapes better than the original and improved seam carving, and can be easily combined with other types of energy functions within the seam carving framework.
- 13
- 13
- 12
- 12
- Metric Compatible Training for Online Backfilling in Large-Scale Retrieval11
This work introduces an online backfilling algorithm, which enables it to achieve a progressive performance improvement during the backfilling process without sacrificing the full performance of the new model after the completion of backfilling.
- 11
- 10
- Map the Flow: Revealing Hidden Pathways of Information in VideoLLMs10
This study investigates the internal information flow of VideoLLMs using mechanistic interpretability techniques and reveals consistent patterns across diverse VideoQA tasks, providing a blueprint for how VideoLLMs perform temporal reasoning and offer practical insights for improving model interpretability and downstream generalization.
- 10
- 10
- Multi-Modal Representation Learning with Text-Driven Soft Masks10
A visual-linguistic representation learning approach within a self-supervised learning framework by introducing a new operation, loss, and data augmentation strategy that encourages the model to focus more on hard but diverse examples, leading to outstanding performance on multiple vision-language downstream tasks.
- Enhanced Diffusion Sampling via Extrapolation with Multiple ODE Solutions9
This work proposes an enhanced ODE-based sampling method for DPMs inspired by Richardson extrapolation, which reduces numerical error and improves convergence rates, and improves the quality of generated samples without requiring additional sampling iterations.
- 9
- A Training-Free Defense Framework for Robust Learned Image Compression8
This work proposes a simple but effective two-way compression algorithm with random input transforms, which is conveniently applicable to existing image compression models and preserves the original rate-distortion performance of the models on clean images.
- 8
- Product Quantized Translation for Fast Nearest Neighbor Search8
The proposed algorithm aims to eliminate nearest neighbor candidates effectively using their distance lower bounds in nonlinear embedded spaces, and further improves the lower bounds by transforming data and query through product quantized translations, which is illustrated quantitatively using various large-scale benchmark datasets in different sizes and dimensions.
- 8
- Diffusion-Based Conditional Image Editing Through Optimized Inference with Guidance6
A simple but effective training-free approach for text-driven image-to-image translation based on a pre-trained text-to-image diffusion model that improves the fidelity of the generated target image to the given target prompt while maintaining the structure integrity of the source image.
- Randomized Adversarial Style Perturbations for Domain Generalization6
The proposed RASP algorithm perturbs the style of a feature in an adversarial direction towards a randomly selected class, and is evaluated via extensive experiments to show that the approach improves domain generalization performance, especially in large-scale benchmarks.
- Efficient Decoupled Neural Architecture Search by Structure And Operation Sampling6
A novel neural architecture search algorithm via reinforcement learning by decoupling structure and operation search is proposed that improves the efficiency of architecture search significantly compared to the existing methods while achieving competitive classification accuracy and model compactness.
- 6
- Variational Distribution Learning for Unsupervised Text-to-Image Generation5
This work employs a pretrained CLIP model, which is capable of properly aligning embeddings of images and corresponding texts in a joint space and, consequently, works well on zero-shot recognition tasks.
- Open-Set Representation Learning through Combinatorial Embedding5
A combinatorial learning approach, which naturally clusters the examples in unseen classes using the compositional knowledge given by multiple supervised meta-classifiers on heterogeneous label spaces, and discovers novel concepts via a joint optimization for enhancing the discrimitiveness of unseen classes as well as learning the representations of known classes generalizable to novel ones.
- 5
- 5
- Dynamic Resource Allocation by Ranking SVM for Particle Filter Tracking5
A dynamic resource allocation algorithm based on Ranking Support Vector Machine (R-SVM) for particle filter tracking that is useful to reduce observation cost and improve sampling quality in particle filtering.
- Beyond the Ground Truth: Enhanced Supervision for Image Restoration4
This work proposes a novel framework that enhances existing ground truth images to provide higher-quality supervision for real-world restoration and generates perceptually enhanced ground truth images using super-resolution by incorporating adaptive frequency masks, which are learned by a conditional frequency mask generator.
- ICM-SR: Image-Conditioned Manifold Regularization for Image Super-Resolution4
The proposed image-conditioned manifold regularization (ICM), a method that regularizes the output towards a manifold conditioned on the sparse yet essential structural information: a combination of colormap and Canny edges, significantly enhances super-resolution performance, particularly in perceptual quality.
- Observation-Guided Diffusion Probabilistic Models4
A novel diffusion-based image generation method called the observation-guided diffusion probabilis-tic model (OGDM), which effectively addresses the trade-off between quality control and fast sampling by integrating the guidance of the observation process with the Markov chain in a principled way.
- Learning with Noisy Labels: Interconnection of Two Expectation-Maximizations4
This work addresses learning with noisy labels (LNL) problem, which is formalized as a task of finding a structured manifold in the midst of noisy data, by providing a proper objective function and an optimization algorithm based on two expectation-maximization (EM) cycles.
- Adaptive Kernel Density Approximation and Its Applications to Real-Time Computer Vision4
This dissertation describes a compact and flexible representation of probability density functions using a mixture of Gaussians which is called Kernel Density Approximation (KDA), and introduces a Kernel-based Bayesian filtering framework, which is applied to object tracking and shows improved performance with a smaller number of samples.
- Human-Centric AI: From Explainability and Trustworthiness to Actionable Ethics3
This workshop will explore key challenges and emerging solutions in the development of human-centric AI, with a focus on explainability, trustworthiness, fairness, and privacy.
- Image Diffusion Preview with Consistency Solver3
Diffusion Preview is introduced, a novel paradigm employing rapid, low-step sampling to generate preliminary outputs for user evaluation, deferring full-step refinement until the preview is deemed satisfactory, making it ideal for efficient preview-and-refine workflows.
- TextGuider: Training-Free Guidance for Text Rendering via Attention Alignment3
This work proposes TextGuider, a novel training-free method that encourages accurate and complete text appearance by aligning textual content tokens and text regions in the image by analyzing attention patterns in Multi-Modal Diffusion Transformer (MM-DiT) models.
- 3
- Rank-based voting with inclusion relationship for accurate image search3
This work presents a rank-based voting technique utilizing inclusion relationship for high-quality image search, and proposes a novel similarity measure based on representative regions and their inclusion relationship.
- Occlusion detection using horizontally segmented windows for vehicle tracking3
Experimental results in real scenes show that the proposed method finds the occluded region successfully and improves the performance of the conventional trackers.
- Macrofeature layout selection for pedestrian localization and its acceleration using GPU3
This work proposes a macrofeature layout selection technique to improve localization performance in an object detection task and employs line, triangle, and pyramid layouts, which are composed of several local blocks represented by the Histograms of Oriented Gradients features in a multi-scale feature pyramid.
- 2
- 2
- 2
- Forget and Diversify: Regularized Refinement for Weakly Supervised Object Detection2
This work proposes simple yet effective regularization techniques, weight reinitialization and labeling perturbations, which prevent overfitting to noisy labels by forgetting biased weights and introduces a graph-based mode-seeking technique that identifies multiple object instances in a principled way.
- Confidence Calibration in Deep Neural Networks through Stochastic Inferences2
A novel variance-weighted confidence-integrated loss function that is composed of two cross-entropy loss terms with respect to ground-truth and uniform distribution, which are balanced by variance of stochastic prediction scores is designed.
- Adaptive Resource Management for Sensor Fusion in Visual Tracking2
This work describes a method to combine information from multiple sensors and estimate the current tracker state by using a mixture of sequential Bayesian filters, and applies this technique to visual tracking problems with multiple cameras or multiple features, and demonstrates its effectiveness through tracking results in real videos.
- 2
- Emergence of Text Readability in Vision Language Models1
The analysis reveals a critical phenomenon: the ability to read textual information in a given image emerges abruptly after substantial training iterations, in contrast to semantic content understanding which develops gradually from the early stages of training.
- FedLPA: Local Prior Alignment for Heterogeneous Federated Generalized Category Discovery1
Local Prior Alignment (LPA) is introduced, a self-distillation mechanism that aligns batch-level predictions with empirical class prior derived from concept assignments that achieves robust generalized category discovery under severe data heterogeneity.
- Re-Evaluating Group Robustness via Adaptive Class-Specific Scaling1
This approach reveals that a naive ERM baseline matches or even outperforms the recent debi-asing methods by simply adopting the class-specific scaling technique, and introduces a novel unified metric that quantifies the trade-off between the two accuracies as a scalar value, allowing for a comprehensive evaluation of existing algorithms.
- 1
- 1
- 1
- 1
- 1
- –
- –
- –
- Merge and Bound: Direct Manipulations on Weights for Class Incremental Learning–
A novel training approach, named Merge-and-Bound (M&B) for Class Incremental Learning (CIL), which directly manipulates model weights in the parameter space for optimization and reveals that it is possible to effectively obtain new models near old ones, reducing catastrophic forgetting.
- STR-Match: Matching SpatioTemporal Relevance Score for Training-Free Video Editing–
STR-Match is a training-free video editing algorithm that produces visually appealing and spatiotemporally coherent videos through latent optimization guided by the novel STR score, which captures spatiotemporal pixel relevance across adjacent frames.
- –
- ContraNeRF: 3D-Aware Generative Model via Contrastive Learning with Unsupervised Implicit Pose Embedding–
A novel 3D-aware GAN optimization technique through contrastive learning with implicit pose embeddings that outperforms existing methods by large margins on the datasets with multiple object categories and inconsistent canonical camera poses.
- –
- Beyond homography: nonparametric image alignment via graph convolutional networks–
A novel nonparametric transformation model based on graph convolutional networks without an explicit geometric constraint is proposed, which is generic and flexible in the sense that it is applicable to the image pairs undergoing diverse local and/or global transformations.
- –
- –
- –
- –
- –
- –
- –
- –
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.