Every paper checked — nothing hidden
All papers
Every paper checked in the week of 12–18 Oct 2026 · Multimodal models — nothing is hidden. Brief papers first.
All1,924In brief3More (on-topic)553Not relevant1,333Not assessed19Older33Could not confirm19Blogs with no paper11All blogs13
| Paper | Topic | Relevance | Feedback |
|---|---|---|---|
| GenIA: Generative Reconstruction with Test-Time Input Alignment Presents a test-time alignment framework that grounds generative 3D foundation models in observed geometry and appearance without model retraining. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Deals with 3D computer vision reconstruction rather than multimodal vision-language interaction. | |
| WorldCast: Distributed Multiplayer World Models Proposes a distributed multiplayer world model architecture where individual clients generate synchronized video views by exchanging player state fields. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on distributed video generation and 3D world models without multimodal cross-modality capabilities. | |
| Beyond Spatio-Temporal Priors: A Generalizable Approach for Dense Correspondence Matching Presents a dense correspondence framework combining generative and semantic representations to handle non-rigid image editing and generation. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Addresses vision-only dense correspondence matching rather than multimodal cross-modality reasoning. | |
| Pumpire: Unified Benchmark for Metric Distance Estimation Introduces a benchmark evaluating metric point-to-point distance estimation in 3D foundation models using physical ground-truth measurements. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses purely on 3D computer vision metric distance estimation without multimodal integration. | |
| One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts Designs a vision transformer using a single recurrent block with depth-programmed experts to reduce stored parameters without accuracy loss. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on single-modality vision transformer architectures rather than multimodal models. | |
| Dex-One2Many: Learning Dexterous Manipulation from a Single Human Demonstration Learns dexterous manipulation policies from a single human video by abstracting motions into sequential scene graphs for reinforcement learning. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on reinforcement learning policy acquisition from demonstration rather than multimodal modeling. | |
| Fly-by-Code: Embodied Coding Agents for Aerial Manipulation with Active Visual and Physical Feedback Presents an embodied coding agent framework using active physical and visual probing to refine aerial manipulation policies. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on embodied robotic control programs rather than multimodal foundational model architectures. | |
| Scaling Multi-Teacher Distillation for Digital Pathology Introduces an architecture with teacher-specific tokens and mixture-of-experts to scale multi-teacher distillation in digital pathology. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on unimodal vision encoder distillation for pathology slides rather than multimodal models. | |
| MemReg: Streaming Outdoor LiDAR Point Cloud Registration with Hybrid Memory Buffers Proposes a multi-frame outdoor LiDAR registration network using hybrid feature and pose memory buffers to leverage temporal sequences. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Deals strictly with 3D LiDAR point clouds without cross-modal interaction. | |
| WoodBR-Cross: A public wood-image dataset and individual-tree-disjoint benchmark for species identification Presents an image dataset and an individual-tree-disjoint evaluation benchmark for automated wood species identification. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses purely on single-modality wood image classification, not multimodal models. | |
| Bridging Human and Model Attention: Explainability Analysis of CNN, Mamba, and ViT Architectures with Gaze-Based Validation Evaluates explainability methods across CNN, ViT, and Mamba models using eye-tracking gaze data as ground truth. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Analyzes vision architectures on image datasets without multimodal cross-modal modeling. | |
| Zebra3D: Efficient Hybrid Foundation Models For Scalable 3D Reconstruction Proposes a hybrid linear and sparse attention framework to accelerate feed-forward 3D foundation models across many views. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Addresses 3D vision reconstruction, not multimodal vision-language reasoning. | |
| Reference-Free Food Volume Estimation from Egocentric Stereo Video Presents an automatic reference-free method to reconstruct and estimate food portion volumes using egocentric stereo video. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Uses standard 3D reconstruction and segmentation methods without multimodal cross-modality learning. | |
| Kernelized Parameter Sensitivity Atlases for Amortized Conditional Interpretability Develops a kernelized parameter sensitivity atlas to amortize input-specific gradient attribution across vision, forecasting, and biology models. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Deals with parameter sensitivity atlases, not multimodal fusion or vision-language models. | |
| Domain Adaptation of SAM 2.1 for Building-Component Segmentation in Image-Based 3D Reconstruction Fine-tunes SAM 2.1 on multi-class building images to improve 2D segmentation for image-based 3D reconstruction. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Applies 2D vision models to 3D radiance fields without multimodal text or audio features. | |
| Structure-Aware Matching for Robust UAV Thermal-to-Satellite Geo-Localization Proposes a structure-aware matching framework exploiting directional and low-frequency patterns for UAV thermal-to-satellite geo-localization. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Deals with cross-view thermal-to-optical matching, not multimodal text-image or cross-domain LLMs. | |
| APEX: Asynchronous Prefix Denoising for Extreme Compression Introduces an asynchronous prefix denoising framework for extreme latent compression in diffusion models. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Applies to image autoencoding and generation, not multimodal vision-language models. | |
| Selective Spatio-Temporal Fruit Inspector (S$^2$FI): A Mamba-based Framework for Rotational Fruit Quality Assessment Proposes a lightweight fruit sorting system combining 3D convolutions and Mamba sequence modeling for rotational visual quality inspection. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses purely on video and 3D vision modeling without language or other modal integration. | |
| ROSETTA: Rotation Subspace Estimation for Training-free Transmodal Alignment Presents a training-free calibration layer that aligns mismatched cross-modal embedding spaces via local rotation estimation. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Addresses sensor coordinate cross-modal alignment without language or multimodal LLM components. | |
| Win-Win Fusion: Multi-Modal Image Fusion with Visual-Semantic Collaboration Develops a multi-modal image fusion framework that combines structural guidance and interactive gated attention to improve perception. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on low-level visual image fusion (sensor fusion), not vision-language foundation models. | |
| MeshBind: Point-to-Mesh Binding for Non-Rigid Scene Flow Automatic Annotation Presents an automatic scene flow annotation pipeline using articulated mesh binding for non-rigid objects in autonomous driving datasets. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Involves 3D vision and LiDAR sensors, not multimodal language or cross-modality reasoning. | |
| AD-SRF: Attack-Defense Semantic Reinforcement for Multi-Modal Image Fusion Proposes an attack-defense game framework to reinforce semantic attributes in multi-modal image fusion while preserving visual quality. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Deals with low-level sensor image fusion, not vision-language or multimodal LLMs. | |
| JMT-YOLO: Multi-task Domain Adaptation via Segmentation-Guided Consistency Learning Proposes a domain adaptive object detection framework using segmentation-guided consistency learning to handle distribution shifts. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Addresses vision-only object detection and segmentation, not multimodal cross-modal modeling. | |
| What Egocentric Video Embeddings Encode and How to Combine Them with Inertial Sensors for Human Activity Recognition Analyzes what egocentric video embeddings encode and demonstrates complementary sensor fusion with wearable inertial measurement units. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Studies egocentric video and inertial sensors, not vision-language or standard multimodal foundation models. | |
| UniDeblur: Unified Motion Deblurring via Fourier Convolution Learning Develops a unified motion deblurring framework using Fourier convolution learning to handle diverse real-world blur kernels. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Deals with unimodal image deblurring and restoration without multimodal components. | |
| Learning Representations to Guide Pixel Diffusion Transformers Investigates internal representation utility in pixel diffusion transformers and proposes Representation Guidance to improve generation performance. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on unimodal image generation and representation learning rather than multimodal models. | |
| Learning Features, Not Layouts: Feature-Axis Attention as a Transfer Prior for Medical Imaging Designs a vision transformer prioritizing feature-axis attention over spatial attention to improve cross-domain transfer in medical imaging. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Explores attention axis design in vision transformers without multimodal or language integration. | |
| FlyHack: Stress-Testing Mechanistic Claims in a Connectome-Constrained Visual Model Stress-tests mechanistic and behavioral claims in a connectome-constrained Drosophila motion-vision neural network model. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Investigates computational neuroscience and biological vision models rather than multimodal AI architectures. | |
| Unified Generative Video Compression with Autoregressive and Bidirectional Video Diffusion Presents a generative video codec supporting both low-delay autoregressive streaming and high-quality bidirectional playback from a single bitstream. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Pertains to video compression with diffusion models without cross-modal multimodal integration. | |
| What Visual Representations Do World Action Models Need? Evaluates visual representations and integration mechanisms for video-based world action models in robot manipulation tasks. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Investigates visual representations for action/manipulation models rather than multimodal text-image-audio integration. | |
| Dataset-Specific Encoding of Image Content Underlies Cross-Dataset Failure in Brain MRI Foundation Models Investigates cross-dataset transfer failures in brain MRI foundation models and attributes them to dataset-specific encoding directions of image content. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Studies visual representations of MRI scans without multimodal or language-grounded components. | |
| NOFACE3D: NOise-scheduled Focused Anchoring and Controlled Enhancement for 3D Generation Develops a training-free framework to refine fine-scale details in generated 3D meshes using focused bounding-box re-encoding. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Pertains to single-image 3D mesh generation rather than multimodal reasoning or vision-language models. | |
| Volume Rendering with Metric--Phase Fields for Thin Structures and Sign-Ambiguous Geometry Introduces PhaseSurf, combining metric-phase fields and volume rendering to reconstruct thin structures and sign-ambiguous geometry. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses entirely on 3D geometric surface reconstruction rather than cross-modal understanding. | |
| Stochastic Finite-Transition Learning and Exploration for Accelerated MRI Reconstruction Accelerates generative MRI reconstruction by learning transitions across discrete k-space intervals combined with stochastic exploration. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Deals with unimodal MRI reconstruction without multimodal or vision-language modeling. | |
| Replaying the Movie in Your Mind Decoding Dynamic Visual Perception and Recall from EEG Releases an EEG benchmark and evaluates decoding models comparing neural representations during video perception versus internal mental recall. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Studies decoding EEG neural signals rather than multimodal vision-language or cross-modal AI models. | |
| Fix the Match, Not the Solver: STITCH -- a Self-Supervised Compatibility Measure for Eroded Visual Puzzles Presents a self-supervised interior patch compatibility measure to solve eroded visual jigsaw puzzles without clean boundaries. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Proposes a patch-matching vision architecture without language or cross-modal integration. | |
| CompactSplat: Spatially Adaptive Gaussian Distribution for Feedforward Scene Reconstruction Introduces a content-aware 3D Gaussian Splatting framework that adaptively distributes Gaussian primitives based on scene texture and structure. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses solely on 3D novel view synthesis and Gaussian Splatting without multimodal elements. | |
| High-Throughput Generative Video Super-Resolution on a Single Consumer GPU Presents a training-free generative video super-resolution pipeline combining one-step diffusion and flow interpolation for consumer GPUs. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses strictly on diffusion-based video super-resolution without multimodal text or language interaction. | |
| Complexity-Calibrated Reconstruction Discrepancy for No-Reference Point Cloud Quality Assessment Presents a reference-free point cloud quality assessment method using successive surface reconstructions to decouple degradation from content complexity. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Studies 3D geometry and 2D projection discrepancy without multimodal language or text integration. | |
| Multi-Modal Graph Neural Network-Inspired Vessel Identification and Tracking for UAV-Based Maritime Surveillance Combines multimodal graph learning and perspective transformation of visible and infrared feeds for drone-based vessel identification and tracking. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Combines sensor data (infrared, AIS, visual) via GNNs rather than text-image multimodal foundation models. | |
| Automated Detection and Encoding of Cartographic Signs in Historical Maps Presents a detection and embedding framework alongside a benchmark dataset for recognizing cartographic signs in historical archival maps. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on object detection and metric learning for historical maps without multimodal language modeling. | |
| FPD-Drive: Future Policy Distillation for End-to-End Autonomous Driving Develops a distillation framework transferring privileged future trajectory knowledge and safety scoring to improve end-to-end autonomous driving planners. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on end-to-end trajectory planning and policy distillation, not multimodal language-vision models. | |
| The Predictability Bias of Joint-Embedding Predictive Learning in Medical Imaging Audits joint-embedding predictive learning in medical CT scans, showing a bias that degrades performance on unpredictable localized pathology. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Studies 3D CT vision encoders (JEPA) without language or general multimodal model integration. | |
| Deep Learning Approach for RealTime Contraction Detection in Laboring Dairy Cattle Applies convolutional neural networks and vision transformers to video feeds for automated detection of labor contractions in dairy cattle. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Uses standard CNNs and ViTs on video frames rather than multimodal reasoning or vision-language integration. | |
| Beyond a Single Crop: Contextualized Embeddings for Player Re-Identification Introduces ContextReID, a framework utilizing co-occurring players and cross-frame sequences to improve player re-identification in broadcast sports video. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Uses a vision-language model solely as an off-the-shelf tool for a narrow sports re-identification task. | |
| QCLR: A Plug-in Robustification Module via Quantization for Contrastive Learning in Self-Supervised Skeleton-Based Human Action Recognition Proposes a quantization module to improve adversarial robustness in self-supervised skeleton-based human action recognition. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Addresses skeleton action recognition without multimodal language or cross-modal components. | |
| Multi-Source and Multi-Dimensional Information Fusion in Resource-Constrained Edge Networks Develops a semantic information fusion framework that coordinates distributed edge terminal uploads and slot-based token aggregation. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on distributed edge communication and token transmission rather than multimodal foundation models. | |
| Decoding Register Tokens Preserves Global Scene Details Investigates decoding register tokens alongside patch tokens in vision transformers to preserve global scene appearance details. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses exclusively on visual patch and register tokens within vision models. | |
| NovaSplat: Model-Free 6D Object Pose Estimation via 3D Gaussian Splatting Onboarding and Refinement Presents a model-free pipeline using smartphone-derived 3D Gaussian Splatting for novel object onboarding and 6D pose estimation. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on 3D computer vision and pose estimation without multimodal reasoning. | |
| CoVA: Composing VAE Representations for Generative Learning Composes internal VAE representations into stage-wise guidance to improve diffusion transformer generative learning without external encoders. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Investigates visual generative model representations without cross-modal multimodal integration. | |
| VideoAWR: Reinforcing Video Flow Models with Decoupled-Timestep Advantage-Weighted Rectification Introduces a reinforcement learning method that decouples sampling from optimization anchors to stabilize video flow generation training. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on generative video flow models rather than multimodal reasoning systems. | |
| Why DiT Models Underperform as Representation Learners without Long Skip Connections Analyzes how long skip connections in Diffusion Transformers preserve feature representations for unified visual generation and understanding. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Studies unimodal visual feature representations inside diffusion transformers. | |
| Radiomics-Based Thyroid Nodule Classification with Label-Conditional Conformal Prediction Combines radiomics features, an additive model, and conformal prediction for uncertainty-aware thyroid nodule ultrasound classification. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on radiomics and tabular classification without multimodal model architectures. | |
| SADL: A Size-Adaptive Decision Layer for Chronic Stroke Lesion Segmentation in ISLES 2026 Introduces an adaptive decision layer and ensemble scheme for chronic stroke lesion segmentation on brain MRI. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses strictly on medical image segmentation without multimodal language-vision capabilities. | |
| A Real-World CAD Segmentation Benchmark: Comparing Rule-Based and Learning-Based Approaches through Hole Detection Introduces a real-world CAD segmentation benchmark focused on fastener hole detection, comparing rule-based approaches with deep learning methods. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Evaluates 3D CAD mesh/B-Rep segmentation without multimodal or vision-language integration. | |
| GraftOOD: Grafting Out-of-Distribution Exemplars into Advantage-Weighted Rectification for Video Reward Optimization Grafts out-of-distribution video exemplars into reward optimization to prevent motion collapse during video generation training. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on video generation dynamics and reward optimization rather than multimodal reasoning. | |
| SplatQ: Interaction-Aware Pruning of 3D Gaussian Splatting via Quadratic Selection Presents a quadratic selection pruning algorithm for 3D Gaussian Splatting that coordinates regional budgets and accounts for joint deletion effects. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses purely on 3D scene reconstruction and graphics pruning without cross-modal interaction. | |
| Beyond Successful Robot Demonstrations: Learning from Human Interactions and Modeling Failure Dynamics Develops a robotic action-conditioned world model trained on egocentric human interactions and failure dynamics to improve execution foresight. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Focuses on robotic action-conditioned world models rather than multimodal reasoning or vision-language models. | |
| Efficient Confidence-Aware Densification for 3DGS Introduces an error-driven, confidence-aware densification method to place primitives more effectively in 3D Gaussian Splatting under fixed budgets. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Deals strictly with 3D novel view synthesis rather than multimodal vision-language architectures. | |
| MOSAIC: Multi-Scale Object-Centric Slot Adaptation with Image-Guided Prototype Conditioning for Remote Sensing Image Super-Resolution Proposes an object-centric slot adaptation framework with prototype memory for remote sensing image super-resolution using diffusion Transformers. Preprint | Multimodal models via arxiv · 12 Oct, 16:28 | Not relevant Deals with image-to-image super-resolution via diffusion Transformers without multimodal integration. |
1,681–1,740 of 1,924← PreviousNext →