8780 papers
This paper proposes MomentMix, a data augmentation strategy combining ForegroundMix and BackgroundMix, and a Length-Aware Decoder to address feature diversity limitations and prediction biases, thereby significantly improving the localization accuracy of short moments in Video Moment Retrieval tasks.
This paper proposes JOSH, an optimization-based method that jointly reconstructs 4D human motion and surrounding scenes from monocular web videos by leveraging human-scene contact constraints, along with its efficient learning-based variant JOSH3R trained on pseudo-labels derived from JOSH.
This paper challenges the prevailing reliance on diffusion models for adversarial defense by demonstrating that non-diffusion purifiers can achieve superior robustness, transferability, and cross-dataset generalization, notably outperforming ImageNet-trained diffusion models when applied to ImageNet despite being trained only on CIFAR-10.
This paper introduces Dual-IPO, an iterative framework that simultaneously and progressively optimizes both a CoT-guided reward model and a video generation model to enhance text-to-video synthesis quality and human preference alignment without requiring extensive manual annotations.
The paper proposes RelaCtrl, a relevance-guided framework that optimizes control signal integration in Diffusion Transformers by dynamically tailoring layer configurations and introducing a Two-Dimensional Shuffle Mixer, achieving superior performance with only 15% of the parameters and computational complexity compared to PixArt-delta.
This paper introduces U-F-CBM, a novel method that transforms any frozen visual classifier into an unsupervised, label-free, and CLIP-free Concept Bottleneck Model by aligning its class distribution with vision-language counterparts, thereby achieving state-of-the-art performance without requiring manual annotations or pre-trained CLIP models.
UniFuture introduces a unified 4D driving world model that jointly generates future RGB images and depth maps through a dual-latent sharing scheme and multi-scale latent interaction, achieving superior performance in both dynamic scene forecasting and geometric perception compared to existing specialized models.
Inspired by the convolution theorem, this paper analyzes gating mechanisms from a frequency perspective to reveal their role in managing frequency responses, leading to the proposal of GmNet, a lightweight model that mitigates low-frequency bias and achieves high performance in image classification.
The paper introduces ViT-Linearizer, a cross-architecture distillation framework that transfers the rich representations of quadratic-complexity Vision Transformers into efficient linear-time recurrent models (such as Mamba) via activation matching and masked prediction, achieving competitive ImageNet accuracy while significantly reducing inference costs for high-resolution tasks.
The paper introduces LAMM-ViT, a novel Vision Transformer that enhances AI face detection by integrating Region-Guided Multi-Head Attention with dynamic Layer-aware Mask Modulation to capture hierarchical structural inconsistencies across diverse generative models, achieving state-of-the-art generalization performance.
This paper proposes a Reflectance Prediction-based Knowledge Distillation (RPKD) framework that enhances 3D object detection robustness in low-bitrate compressed point clouds by discarding reflectance during transmission, reconstructing it via a geometry-based prediction module, and utilizing a cross-source distillation strategy to transfer knowledge from raw to compressed data.
BriGeS is a resource-efficient method for generalized monocular depth estimation that fuses geometric and semantic foundation models via a trainable Bridging Gate and Attention Temperature Scaling to achieve state-of-the-art performance in complex scenes.
This paper proposes "Sparse Imagination," a transformer-based visual world model planning method that utilizes a randomized grouped attention strategy to dynamically reduce token processing during latent rollout, thereby significantly accelerating inference efficiency while maintaining high control fidelity for real-time robotic applications.
LinGuinE is a novel, training-free PyTorch framework that achieves state-of-the-art longitudinal volumetric tumour segmentation and lesion tracking across multiple datasets by combining image registration with guided segmentation from a single radiologist prompt, enabling flexible, direction-agnostic analysis without requiring longitudinal data training.
This paper proposes a novel human-guided framework for CBCT-to-MDCT translation that leverages a Schrödinger Bridge formulation with conditional diffusion and classifier-free guidance to effectively suppress shade artifacts while preserving anatomical fidelity and aligning with clinical preferences through iterative human feedback.
This paper addresses the "Data Addition Dilemma" in medical image segmentation by proposing an exchangeability-based framework that controls foreground-background feature discrepancies across deep network layers, achieving state-of-the-art performance on five datasets including a novel curated ultrasound collection.
LayerT2V is a unified framework that generates semantically consistent, editable multi-layer videos (including background, foregrounds, and alpha mattes) in a single inference pass by leveraging temporal serialization within a shared DiT backbone, supported by the new large-scale VidLayer dataset.
RAP is a unified framework that enables real-time, high-quality audio-driven portrait animation by introducing a hybrid attention mechanism for fine-grained audio control and a static-dynamic training-inference paradigm to overcome the limitations of compressed latent representations.
This paper proposes MixCache, a training-free framework that accelerates video DiT inference by employing a context-aware triggering mechanism and an adaptive hybrid strategy to dynamically select optimal caching granularities, thereby significantly improving both generation speed and quality.
This paper introduces Dyslexify, a training-free defense mechanism that selectively ablates specific attention heads in CLIP vision encoders to neutralize typographic attacks, significantly improving robustness against text-based manipulations while preserving standard recognition accuracy.