Every paper checked — nothing hidden

All papers

Every paper checked for brief #4 of 12–19 Oct 2026 — nothing is hidden. Brief papers first.

All17,313 In brief24 More (on-topic)3,333 Not relevant13,754 Not assessed87 Older352 Could not confirm93 Blogs with no paper266 All blogs284

Paper	Topic	Relevance
Auditing VLM Benchmarks from Existing Evaluation Runs Audits vision-language model benchmarks by analyzing saved evaluation artifacts and removing image inputs to detect ungrounded questions. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Evaluates general vision-language benchmarks, not industrial manufacturing or inspection computer vision.
ELMA: Benchmarking Anaphoric Compositions for Long-Term Text-to-Motion Generation Introduces ELMA, a benchmark dataset for generating long-term human motions from multi-segment instructions with anaphoric references. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on generating avatar human motions, not industrial computer vision or robotic inspection.
Learning Holistic Scene Layout against View Shifts in Cross-View Geo-Localization Applies multi-directional spatial context modeling and high-frequency guidance for cross-view geographic localization. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on geographic cross-view localization, not industrial vision, manufacturing, or defect detection.
Learning Holistic Scene Layout against View Shifts in Cross-View Geo-Localization Applies multi-directional spatial context modeling and high-frequency guidance for cross-view geographic localization. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Addresses visual geo-localization using image features, lacking multimodal integration.
What We Talk About When We Talk About Emotion Editing Uses fMRI encoding representations to guide targeted emotional edits on images while preserving visual content. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on neural-guided image generation and editing rather than multimodal language-vision architectures.
What We Talk About When We Talk About Emotion Editing Uses fMRI encoding representations to guide targeted emotional edits on images while preserving visual content. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Explores emotional image editing via fMRI signals, not industrial inspection or defect detection.
RARA: Robust Policy Fine-Tuning through Retrieval-Anchored Representation Alignment for Robotic Manipulation Aligns robot policy representations with retrieved pretraining examples to maintain generalization during low-data fine-tuning. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Focuses on policy fine-tuning techniques rather than LLM benchmarks or evaluation methods.
RARA: Robust Policy Fine-Tuning through Retrieval-Anchored Representation Alignment for Robotic Manipulation Aligns robot policy representations with retrieved pretraining examples to maintain generalization during low-data fine-tuning. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Does not evaluate safety, jailbreaks, hallucination, or operational guardrails.
RARA: Robust Policy Fine-Tuning through Retrieval-Anchored Representation Alignment for Robotic Manipulation Aligns robot policy representations with retrieved pretraining examples to maintain generalization during low-data fine-tuning. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses out-of-distribution policy adaptation rather than model quantization, distillation, or efficient serving.
Thyroid Nodule Segmentation in Ultrasound Images with Conformal Uncertainty Estimation Combines deep segmentation networks with conformal prediction to estimate pixel-level uncertainty on thyroid ultrasound images. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Covers unimodal medical image segmentation, containing no multimodal integration.
Thyroid Nodule Segmentation in Ultrasound Images with Conformal Uncertainty Estimation Combines deep segmentation networks with conformal prediction to estimate pixel-level uncertainty on thyroid ultrasound images. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Investigates medical thyroid ultrasound segmentation, not industrial manufacturing or visual inspection.
CUBE: Class-conditioned Unknown Boundary Evidence for Open-Set Insect Species Recognition Develops a class-conditioned boundary evidence method for open-set recognition of visually similar insect species. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on biodiversity insect species classification, not industrial computer vision or defect detection.
CUBE: Class-conditioned Unknown Boundary Evidence for Open-Set Insect Species Recognition Develops a class-conditioned boundary evidence method for open-set recognition of visually similar insect species. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on open-set visual classification rather than multimodal vision-language integration.
Benchmarking Answer-Grounding Consistency in Remote Sensing Visual Question Answering Introduces an evaluation protocol and dataset to measure consistency between textual answers and visual grounding in remote sensing VQA. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Addresses remote sensing visual question answering, not industrial defect detection or factory computer vision.
Can Multimodal Decision Models Classify Breast Ultrasound Images? Benchmarking CLEF and CLEF-Flash Across Prompts and Datasets Benchmarks multimodal decision models across multiple prompts for classifying breast ultrasound imaging datasets. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on medical ultrasound diagnostics rather than industrial inspection or factory defect detection.
PhysForge: Material-Controllable 3D Reconstruction with Separable Components Reconstructs physical 3D objects with controllable material properties and interaction dynamics via video diffusion models. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on physical 3D asset generation via video diffusion models rather than multimodal reasoning.
PhysForge: Material-Controllable 3D Reconstruction with Separable Components Reconstructs physical 3D objects with controllable material properties and interaction dynamics via video diffusion models. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on 3D dynamic asset generation with physics simulation, not industrial inspection computer vision.
X-AVDD: Cross-Attentive Audio-Visual Dataset Distillation Proposes a memory-efficient cross-attention method for distilling large paired audio-visual datasets into compact synthetic sets. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Addresses audio-visual dataset distillation, not industrial inspection, defect detection, or factory perception.
X-AVDD: Cross-Attentive Audio-Visual Dataset Distillation Proposes a memory-efficient cross-attention method for distilling large paired audio-visual datasets into compact synthetic sets. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Does not address AI safety, jailbreaks, hallucination, or deployment guardrails.
X-AVDD: Cross-Attentive Audio-Visual Dataset Distillation Proposes a memory-efficient cross-attention method for distilling large paired audio-visual datasets into compact synthetic sets. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Focuses on synthetic dataset distillation for training efficiency rather than evaluating or benchmarking LLMs.
Improving Semantic Coverage in AI Test & Evaluation through Generative Augmentation Uses generative models and vision-language labeling to detect and synthesize missing semantic domains in test datasets. Peer-reviewed	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on general AI evaluation test-set augmentation, not industrial inspection or manufacturing computer vision.
GLIMPSE: Global-Local Indexing with Multi-Precision Spectral Encoding for Explorable Video World Models Designs a global-local memory interface with multi-precision spectral encoding for consistent video world models. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on video world model rendering and memory encoding rather than multimodal vision-language architectures.
GLIMPSE: Global-Local Indexing with Multi-Precision Spectral Encoding for Explorable Video World Models Designs a global-local memory interface with multi-precision spectral encoding for consistent video world models. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Studies memory representations for video world models, not industrial vision or robotics inspection.
State-Isolated Vision Mamba via Update-Tendency Routing for Underwater Image Enhancement Introduces a state-isolation routing mechanism for Vision Mamba models to enhance spatially heterogeneous underwater images. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses solely on low-level image restoration with Vision Mamba, lacking cross-modal multimodal integration.
State-Isolated Vision Mamba via Update-Tendency Routing for Underwater Image Enhancement Introduces a state-isolation routing mechanism for Vision Mamba models to enhance spatially heterogeneous underwater images. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Proposes general underwater image enhancement architecture, not industrial defect detection or factory inspection.
KVeil: First-Place Solution to ECCV 2026 LifeGenIP Challenge on Unlearnable Videos Presents perturbation techniques to protect personal video data from unauthorized identity replication by generative models. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Studies facial privacy protection against video generation models, with no industrial vision application.
KVeil: First-Place Solution to ECCV 2026 LifeGenIP Challenge on Unlearnable Videos Presents perturbation techniques to protect personal video data from unauthorized identity replication by generative models. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on adversarial perturbations against video generation models rather than multimodal understanding or reasoning.
Gaussian Splatting with Reflections Guided by What is Seen Improves 2D Gaussian Splatting rendering of near-field reflections using observed image environment data. Peer-reviewed	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Proposes a general novel view synthesis technique, not industrial inspection or manufacturing computer vision.
Gaussian Splatting with Reflections Guided by What is Seen Improves 2D Gaussian Splatting rendering of near-field reflections using observed image environment data. Peer-reviewed	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses strictly on 3D novel view synthesis with 2D Gaussian Splatting, not multimodal models.
Neurons++: A Unified Visual-Stream Framework for Image and Video Reconstruction from fMRI with Neural-to-Visual Interpretability Reconstructs images and videos from fMRI signals using cortex-inspired serial decoding and multimodal pre-training. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on brain fMRI visual decoding, not industrial computer vision applications.
Neurons++: A Unified Visual-Stream Framework for Image and Video Reconstruction from fMRI with Neural-to-Visual Interpretability Reconstructs images and videos from fMRI signals using cortex-inspired serial decoding and multimodal pre-training. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on brain fMRI-to-vision reconstruction rather than general multimodal language-vision architectures.
Shortcut Learning in Geospatial Foundation Model Embeddings for Cross-Disaster Damage Assessment Evaluates shortcut learning and out-of-distribution failure in geospatial foundation models on cross-disaster damage assessment tasks. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Studies disaster damage assessment using geospatial models, not industrial inspection.
Shortcut Learning in Geospatial Foundation Model Embeddings for Cross-Disaster Damage Assessment Evaluates shortcut learning and out-of-distribution failure in geospatial foundation models on cross-disaster damage assessment tasks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies shortcut learning in remote sensing, not small model optimization.
Shortcut Learning in Geospatial Foundation Model Embeddings for Cross-Disaster Damage Assessment Evaluates shortcut learning and out-of-distribution failure in geospatial foundation models on cross-disaster damage assessment tasks. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Evaluates geospatial foundation models on disaster datasets, not LLM evaluation and benchmarking.
Shortcut Learning in Geospatial Foundation Model Embeddings for Cross-Disaster Damage Assessment Evaluates shortcut learning and out-of-distribution failure in geospatial foundation models on cross-disaster damage assessment tasks. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on satellite imagery embeddings rather than multimodal text-image reasoning.
READ: Reading the Selective Scan for Faithful Video Attribution Provides an attribution method for selective state-space video models by tracing accumulated recurrence in a single forward pass. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Proposes an attribution method for selective state-space video models, not industrial vision.
READ: Reading the Selective Scan for Faithful Video Attribution Provides an attribution method for selective state-space video models by tracing accumulated recurrence in a single forward pass. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on video state-space model attribution rather than multimodal models.
Detection of extremely rare dead cells in fluorescence microscopy images Benchmarks heuristic, machine learning, and deep learning methods for identifying rare dead cells in fluorescence microscopy. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on biological fluorescence microscopy for dead cell screening, not industrial settings.
Detection of extremely rare dead cells in fluorescence microscopy images Benchmarks heuristic, machine learning, and deep learning methods for identifying rare dead cells in fluorescence microscopy. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Deals with biomedical image anomaly detection, not multimodal models.
SimVTA: Simulation-Grounded Video-Trajectory Alignment for Driving World Action Models Proposes a simulation-grounded framework rendering real-world 3DGS scenes to supervise autonomous driving world action models. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on driving trajectory optimization with simulation, not multimodal language-vision integration.
Spanning Diffusion and Flow-based Posterior Sampling via Exact Gaussian Conditioning Presents a unified plug-and-play posterior sampling framework combining diffusion and flow models via exact Gaussian conditioning. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on mathematical posterior sampling for linear inverse problems, not industrial vision.
Spanning Diffusion and Flow-based Posterior Sampling via Exact Gaussian Conditioning Presents a unified plug-and-play posterior sampling framework combining diffusion and flow models via exact Gaussian conditioning. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies trajectory priors for diffusion models, not model compression or efficient serving.
Spanning Diffusion and Flow-based Posterior Sampling via Exact Gaussian Conditioning Presents a unified plug-and-play posterior sampling framework combining diffusion and flow models via exact Gaussian conditioning. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Presents a sampling method for inverse problems, not LLM benchmark evaluation.
Spanning Diffusion and Flow-based Posterior Sampling via Exact Gaussian Conditioning Presents a unified plug-and-play posterior sampling framework combining diffusion and flow models via exact Gaussian conditioning. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Deals with sampling algorithms for diffusion models rather than multimodal reasoning.
Spanning Diffusion and Flow-based Posterior Sampling via Exact Gaussian Conditioning Presents a unified plug-and-play posterior sampling framework combining diffusion and flow models via exact Gaussian conditioning. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on inverse problem solvers rather than LLM safety, red-teaming, or guardrails.
Dream to Remember: Continual Steering of Generalist Robot Policies with World Models Uses a frozen world model to continually steer generalist robot policies onto new tasks without catastrophic forgetting. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses policy steering via world models, not small model distillation or quantization.
Dream to Remember: Continual Steering of Generalist Robot Policies with World Models Uses a frozen world model to continually steer generalist robot policies onto new tasks without catastrophic forgetting. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Investigates continual reinforcement learning for robotics, not AI safety guardrails or reliability.
Dream to Remember: Continual Steering of Generalist Robot Policies with World Models Uses a frozen world model to continually steer generalist robot policies onto new tasks without catastrophic forgetting. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on robot control policy steering rather than multimodal understanding.
Dream to Remember: Continual Steering of Generalist Robot Policies with World Models Uses a frozen world model to continually steer generalist robot policies onto new tasks without catastrophic forgetting. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Focuses on continual robot policy steering rather than evaluating general LLM benchmarks.
Event-based Action Recognition by Pre-Training on Multi-Frame Representations Pre-trains vision transformers on multi-frame event camera data using masked auto-encoding for action recognition. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Applies masked autoencoding to event-camera video representations, not multimodal models.
Event-based Action Recognition by Pre-Training on Multi-Frame Representations Pre-trains vision transformers on multi-frame event camera data using masked auto-encoding for action recognition. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on general action recognition using event cameras, not industrial inspection or manufacturing.
Revisiting the Task of Building Damage Assessment by Satellite Imaging Reviews satellite-based building damage assessment across data, evaluation protocols, and method formulations. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Surveys building damage assessment via satellite imagery for disaster response, not industrial applications.
Revisiting the Task of Building Damage Assessment by Satellite Imaging Reviews satellite-based building damage assessment across data, evaluation protocols, and method formulations. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Examines satellite computer vision benchmarks rather than multimodal text-vision interaction.
LapMerge: Differentiable Exact-Budget Token Merging in Linear Time Introduces a differentiable token merging operator running in linear time with an exact sequence output budget. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Proposes a general token merging operator for sequence models, not industrial vision.
Make It Flow: Animating Water in 3D Gaussian Splatting Scenes Animates flowing water in static 3D Gaussian Splatting scenes using 2D optical flow lifted into a 3D Eulerian motion field. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on animating water in 3D Gaussian Splatting for novel view synthesis.
Make It Flow: Animating Water in 3D Gaussian Splatting Scenes Animates flowing water in static 3D Gaussian Splatting scenes using 2D optical flow lifted into a 3D Eulerian motion field. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Deals with graphics rendering and 3D visual animation rather than multimodal understanding.
Tracking the Invisible: Self-Supervised Wheel Track Detection with Thermal-RGB Fusion Uses self-supervised thermal-RGB fusion to detect drivable wheel tracks on snow-covered roads for autonomous vehicles. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Deals with RGB-thermal fusion for tracking, not broad multimodal vision-language reasoning.
Beyond the Snowflakes: Disentangling Weather Effects from Confounding Meta-Factors in LiDAR Object Detection Analyzes LiDAR object detection in snow to separate atmospheric noise from performance gains caused by increased point reflections. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Analyzes LiDAR object detection robustness rather than multimodal vision-language integration.
Where to Look Matters: Rethinking Sub-Volume Sampling in 3D Medical Self-supervised Learning Replaces random crop sampling in 3D medical self-supervised learning with a guided probe bank based on anatomical priors. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Investigates medical image volume sampling rather than multimodal cross-domain models.
Where to Look Matters: Rethinking Sub-Volume Sampling in 3D Medical Self-supervised Learning Replaces random crop sampling in 3D medical self-supervised learning with a guided probe bank based on anatomical priors. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on 3D medical imaging self-supervised learning, outside industrial inspection and manufacturing.

15,601–15,660 of 17,313← Previous Next →