7585 papers
FedVG is a novel gradient-guided aggregation framework for Federated Learning that utilizes a global validation set to compute layerwise gradient norms, enabling adaptive client weighting that mitigates client drift and enhances model generalization in heterogeneous data settings.
FlowFixer is a refinement framework for subject-driven generation that utilizes a one-step denoising scheme for self-supervised training and direct image-to-image translation to restore fine details lost during scale and perspective changes, thereby outperforming state-of-the-art methods in high-fidelity generation.
This paper introduces PhysicEdit, a new framework that leverages the PhysicTran38K dataset and a textual-visual dual-thinking mechanism to reformulate instruction-based image editing as predictive physical state transitions, thereby significantly improving the physical realism and causal accuracy of generated edits compared to existing methods.
SemVideo introduces a novel fMRI-to-video reconstruction framework that leverages a hierarchical semantic guidance module (SemMiner) to generate multi-level semantic cues, thereby achieving superior temporal coherence and visual consistency compared to existing methods.
This paper proposes Association DETR, a novel object detection model that addresses the limitation of existing SOTA detectors by leveraging neglected background contextual information to achieve state-of-the-art performance on the COCO val2017 dataset.
This paper proposes GFRRN, a novel single image reflection removal network that addresses semantic gaps and label inconsistencies through parameter-efficient fine-tuning, a unified label generator, and advanced frequency and attention mechanisms to achieve state-of-the-art performance.
This paper introduces SPATIALALIGN, a self-improvement framework that enhances text-to-video generation by fine-tuning models with a zeroth-order regularized Direct Preference Optimization method and a novel geometry-based DSR-SCORE metric to better align generated videos with dynamic spatial relationships specified in text prompts.
The paper proposes WARM-CAT, a novel approach for Compositional Zero-Shot Learning that enhances test-time performance by warm-starting a dynamic priority queue and adaptively updating multimodal prototypes through comprehensive knowledge accumulation from unsupervised data, while also introducing refined benchmarks to achieve state-of-the-art results.
This paper introduces a lightweight, real-time framework for motion-aware event suppression that jointly segments and predicts the future motion of independent moving objects and ego-motion to filter dynamic events, achieving state-of-the-art performance on the EVIMO benchmark while significantly accelerating downstream applications like Vision Transformers and visual odometry.
This paper presents a comprehensive theoretical derivation of a unified analytical solution for the spatiotemporal acoustic pressure generated by spherically symmetric photoacoustic sources, providing specific expressions for various initial pressure distributions and accompanying open-source code for ultrafast forward simulation.
This paper introduces AV-LMMDetect, a supervised fine-tuned large multimodal model built on Qwen 2.5 Omni that addresses the generalization limitations of existing detectors by casting audio-video deepfake detection as a prompted classification task, achieving state-of-the-art performance on key datasets.
This paper presents a large-scale empirical study demonstrating that the design of a robotic manipulation policy's action space significantly impacts learning performance, revealing that predicting delta actions improves results while joint-space and task-space representations offer complementary benefits for stability and generalization, respectively.
This paper introduces DesignSense-10k, a large-scale human-annotated dataset and a specialized reward modeling framework that significantly outperforms existing models in evaluating graphic layout aesthetics and enhances downstream layout generation quality through reinforcement learning and inference-time scaling.
SALIENT is a mask-conditioned, wavelet-domain diffusion framework that synthesizes paired lesion-masking volumes to enable controllable, computationally efficient augmentation, thereby significantly improving precision in long-tail rare lesion detection for whole-body CT scans.
This paper characterizes the variety of compatible fundamental matrix triples by computing its multidegree and multihomogeneous vanishing ideal, thereby providing a complete set of algebraic constraints—including a new discovery of simple quartic equations—that improve upon previous incomplete results and also locally define the variety for compatible essential matrix triples.
This paper proposes Structurally-Guided Dynamic Convolution (SGDC), a novel mechanism that replaces traditional pooling-based kernel generation with an explicitly supervised structure-extraction branch to preserve high-frequency boundary details, thereby achieving state-of-the-art performance and superior structural fidelity in medical image segmentation across multiple datasets.
The paper introduces SegReg, a latent-space regularization framework compatible with U-Net architectures that enhances domain generalization and continual learning performance in medical image segmentation by structuring feature embeddings without adding parameters or memory overhead.
To address the scarcity of Dynamic Vision Sensor data, this paper introduces ANTShapes, a Unity-based simulation framework that generates customizable neuromorphic datasets with statistically labeled anomalies for training event-based computer vision models.
This paper proposes a unified proactive forensics framework that employs an innovative 152-dimensional landmark-identity watermark (LIDMark) and a Factorized-Head Decoder to simultaneously achieve robust deepfake detection, tampering localization, and source tracing within a single, imperceptible system.
This paper proposes a few-shot continual learning framework for 3D brain MRI that combines a frozen foundation model with task-specific Low-Rank Adaptation (LoRA) modules to effectively handle sequential tumor segmentation and brain age estimation tasks without catastrophic forgetting, while using fewer than 0.1% trainable parameters per task.