5527 papers
This paper proposes a two-stage method that combines a VQ-VAE for compact latent space embedding with a Structural Causal Model using a closed-form Generalized Linear Model to generate high-quality, diverse 3D brain MRI counterfactuals, effectively overcoming the limitations of existing deep learning and causal models in handling small sample sizes and high-dimensional data.
The paper proposes MV-Adapter, a module incorporating an adaptive channel attention mechanism to enhance the USIS-SAM model's performance in underwater instance segmentation by dynamically adjusting feature weights to mitigate challenges like light attenuation and color distortion.
XPoint is a novel self-supervised, modular framework leveraging a VMamba encoder and joint decoder heads to achieve robust, adaptable multispectral image registration across diverse spectral modalities without relying on expensive labeled data.
EchoMimicV2 is a simplified, high-quality half-body human animation method that leverages an Audio-Pose Dynamic Harmonization strategy, Head Partial Attention for data augmentation, and Phase-specific Denoising Loss to overcome existing limitations in condition redundancy and head-only driving.
This paper presents MFP3D, a novel framework that leverages a single monocular image to generate 3D point clouds and extract combined features for accurate food portion and energy estimation, thereby overcoming the limitations of existing methods that rely on reference objects, depth sensors, or multi-view inputs.
This paper introduces "grounding-IQA," a novel paradigm that integrates multimodal referring and grounding with image quality assessment to enable fine-grained quality perception, supported by the newly constructed GIQA-160K dataset and the GIQA-Bench benchmark.
This paper introduces DAWN-FM, a novel Flow Matching framework that incorporates explicit data and noise embeddings to robustly solve ill-posed inverse problems by learning a time-dependent velocity field for accurate reconstruction and uncertainty quantification.
FiLo++ is a zero-/few-shot anomaly detection framework that enhances performance by generating task-specific fine-grained descriptions via large language models and employing a deformable localization module with multi-scale cross-modal interaction to accurately detect and localize anomalies of varying shapes and sizes.
This paper introduces a family of polynomial, trigonometric, and tropical activation functions derived from orthonormal bases that, when combined with variance-preserving initialization, enable the successful training of deep models like GPT-2 and ConvNeXt while mitigating gradient instability, offering polynomial interpretability, and facilitating fine-tuning through Hermite interpolation.
This study demonstrates that deep generative inference applied to routine CT angiography can accurately localize neural substrates of stroke deficits by analyzing computed perfusion maps, offering a novel, lesion-independent method for early phenotyping and understanding functional anatomical relations in acute ischaemic stroke.
This paper demonstrates that while CLIP's cross-modal alignment fails to preserve compositional structure, causing it to behave like a bag-of-words model, the necessary attribute-object binding information is actually encoded within its unimodal embeddings and can be effectively recovered through a simple linear transformation.
The paper introduces WorldSense, the first benchmark designed to evaluate multimodal LLMs' real-world understanding by integrating synchronized audio, visual, and text inputs across diverse scenarios, revealing that current state-of-the-art models still face significant challenges in achieving coherent omni-modal perception.
This paper demonstrates that less than 1% of a diffusion model's parameters, specifically within attention layers, are responsible for textual content generation, enabling efficient fine-tuning, text editing, and toxic text prevention across diverse model architectures through targeted localization.
This paper proposes a robust adaptation framework for Large Multimodal Models that significantly enhances hateful meme detection performance, cross-domain generalization, and interpretability while outperforming existing supervised fine-tuning and agentic systems.
This paper introduces "Jumbo," a fast and accurate plain Vision Transformer that enhances efficiency by replacing narrow patch tokens with a single, wider, parameter-shared global token processed by an attention-only FFN, thereby improving performance across various tasks while maintaining compatibility with existing ViT methods.
This paper introduces MemeXplain, a new large-scale explanation-enhanced dataset for Arabic propagandistic and English hateful memes, along with a multi-stage optimization strategy for Vision-Language Models that significantly outperforms current state-of-the-art methods in both label detection and rationale generation.
To address the lack of benchmarks for outdoor aerial Vision-Language Navigation, this paper introduces OpenFly, a comprehensive platform featuring a multi-engine simulation environment, an automated data collection toolchain, a large-scale 100k-trajectory dataset, and a specialized keyframe-aware agent model.
The paper introduces LLaVE, a series of large language and vision embedding models that utilize hardness-weighted contrastive learning to overcome the limitations of standard InfoNCE loss, achieving state-of-the-art performance on the MMEB benchmark with superior scalability and zero-shot generalization to video retrieval tasks.
Inspired by DeepSeek-R1, the authors propose Vision-R1, a multimodal large language model that leverages a self-constructed 200K reasoning dataset and a Progressive Thinking Suppression Training strategy with GRPO to significantly enhance multimodal reasoning capabilities, achieving performance comparable to OpenAI O1 on benchmarks like MathVista.
SemHiTok introduces a unified image tokenizer that employs a novel semantic-guided hierarchical codebook to decouple semantic and pixel representations, thereby achieving an optimal trade-off between multimodal understanding and generation tasks.