DataComp-VLM benchmark shows instruction-heavy data mixing outperforms filtering for VLM training, with DCVLM-Baseline achieving 63.6% on 33 tasks for 8B models (+5.4pp over FineVision).
super hub Mixed citations
DINOv2: Learning Robust Visual Features without Supervision
Mixed citation behavior. Most common role is background (44%).
849
Pith papers citing it
Background
44% of classified citations
abstract
The recent breakthroughs in natural language processing for model pretraining on large quantities of data have opened the way for similar foundation models in computer vision. These models could greatly simplify the use of images in any system by producing all-purpose visual features, i.e., features that work across image distributions and tasks without finetuning. This work shows that existing pretraining methods, especially self-supervised methods, can produce such features if trained on enough curated data from diverse sources. We revisit existing approaches and combine different techniques to scale our pretraining in terms of data and model size. Most of the technical contributions aim at accelerating and stabilizing the training at scale. In terms of data, we propose an automatic pipeline to build a dedicated, diverse, and curated image dataset instead of uncurated data, as typically done in the self-supervised literature. In terms of models, we train a ViT model (Dosovitskiy et al., 2020) with 1B parameters and distill it into a series of smaller models that surpass the best available all-purpose features, OpenCLIP (Ilharco et al., 2021) on most of the benchmarks at image and pixel levels.
hub tools
citation-role summary
method 59
background 57
baseline 9
dataset 3
other 1
citation-polarity summary
claims ledger
- abstract The recent breakthroughs in natural language processing for model pretraining on large quantities of data have opened the way for similar foundation models in computer vision. These models could greatly simplify the use of images in any system by producing all-purpose visual features, i.e., features that work across image distributions and tasks without finetuning. This work shows that existing pretraining methods, especially self-supervised methods, can produce such features if trained on enough curated data from diverse sources. We revisit existing approaches and combine different techniques
authors
co-cited works
representative citing papers
WildBox provides over 237k 3D wildlife annotations from drone video and benchmarks reveal zero-shot 3D detection at 0 AP but fine-tuned performance of 8.68 AP-BEV and 13.17 AP3D, with depth estimation causing most errors.
X-Palm supplies the first paired multispectral-to-smartphone palmprint dataset with broad real-world variability to support cross-domain biometric authentication.
Every9D-21M supplies 21.8M real-world 9D pose annotations for 700 everyday categories by propagating manual canonical poses through cross-instance alignment in object-centric videos and verifying them multiview.
A multi-view transformer predicts dense perspective fields that feed a geometric optimizer to estimate camera intrinsics and gravity from arbitrary numbers of real-world views.
Rigel3D jointly generates rigged 3D meshes with geometry, skeleton topology, joint positions, and skinning weights using coupled surface and skeleton latent representations for image-conditioned animation-ready asset synthesis.
Image-to-3D models successfully generate harmful geometries in most cases with under 0.3% caught by commercial filters; existing safeguards are weak but a stacked defense cuts harmful outputs to under 1% at 11% false-positive cost.
neuralCAD-Edit benchmark shows even the best foundation model (GPT 5.2) scores 53% lower than human CAD experts in acceptance trials for multimodal-instructed 3D model edits.
The work creates the first dataset and baseline for generating emission textures on 3D objects to reproduce glowing materials from input images.
Molmo VLMs trained on newly collected PixMo open datasets achieve state-of-the-art performance among open-weight models and surpass multiple proprietary VLMs including Claude 3.5 Sonnet and Gemini 1.5 Pro.
MMMU-Pro is a stricter multimodal benchmark that removes text-only solvable questions, augments options, and requires reading text from images, yielding substantially lower model scores of 16.8-26.9%.
Promptbreeder evolves both task prompts and the mutation prompts that improve them using LLMs, outperforming Chain-of-Thought and Plan-and-Solve on arithmetic and commonsense reasoning benchmarks.
Introduces VICIS task and training framework for inferring visual concepts from image sets, with experiments showing better accuracy, diversity, and generalization than standard VLMs on synthetic and ImageNet data.
InvSplat is a feed-forward multi-view model that predicts 3D Gaussians augmented with intrinsic material attributes for inverse rendering and relighting.
The subspace intervention framework reveals that pre-training objectives shape how ViTs encode geometric information in compressible low-rank subspaces, with peak precision at intermediate layers.
A systematic survey unifies presentation, digital injection, and GenAI synthesis attacks on identity documents, audits datasets for a reality gap, identifies SDGI in multimodal models, and reports APCER above 25% for top models on synthetic IDs.
A training-free prototype memory-guided framework for multi-class prenatal ultrasound anomaly classification and localization using few reference images per class, validated on a 9-category multi-center dataset.
EPO is a trackless, edge-map-alignment framework that refines pose estimates from 3D foundation models and matches or exceeds bundle-adjustment performance with substantially lower runtime and memory use.
GEAR jointly trains VQ tokenizer and AR generator end-to-end via dual hard/soft read-out and representation alignment, achieving up to 10x faster ImageNet gFID convergence than LlamaGen-REPA while generalizing across quantizers and to text-to-image.
WarpHammer densifies scene warps with 3D object priors from generative models and fuses pose-unknown auxiliary views via multi-view geometry to enable stable extreme novel view synthesis.
AnyMatch synthesizes large-scale geometrically consistent multi-modal image pairs from single-view images, enabling fine-tuned matching networks to achieve substantial gains on benchmarks.
A new dataset of 220k+ cross-view pairs and a single-stage geometry-aware model GAGeo based on the π³ 3D foundation model outperforms prior methods on object geo-localization with strong generalization and zero-shot ground-to-drone capability.
First complete digital unwrapping and reading of a Herculaneum papyrus scroll (PHerc. 1667) via synchrotron X-ray CT, virtual unrolling, and machine learning.
Uni-Mo generates 7,488 language-annotated quadruped motions via LLM prompts and video diffusion, lifts them to 3D trajectories, and trains policies achieving 96.7% real-robot success on 392 sampled motions.
citing papers explorer
-
Beyond Voxel 3D Editing: Learning from 3D Masks and Self-Constructed Data
BVE framework enables text-guided 3D editing beyond voxel limits by combining self-constructed data, lightweight semantic injection, and annotation-free masking to preserve local invariance.
-
What Are We Really Measuring? Rethinking Dataset Bias in Web-Scale Natural Image Collections via Unsupervised Semantic Clustering
Supervised dataset discrimination overstates semantic bias via resolution artifacts; unsupervised clustering of foundation-model features on web-scale datasets yields near-chance separability.
-
Boosting Visual Instruction Tuning with Self-Supervised Guidance
Mixing 3-10% of visually grounded self-supervised instructions into visual instruction tuning consistently boosts MLLM performance on vision-centric benchmarks.
-
Pi-HOC: Pairwise 3D Human-Object Contact Estimation
Pi-HOC predicts dense 3D semantic contacts for all human-object pairs in an image via instance-aware tokens and an InteractionFormer, achieving higher accuracy and 20x throughput than prior methods.
-
Relaxing Anchor-Frame Dominance for Mitigating Hallucinations in Video Large Language Models
Decoder-side Temporal Rebalancing (DTR) reduces hallucinations in Video-LLMs by mitigating over-dominance of a single anchor frame during inference without training or auxiliary models.
-
Cross-Attentive Multiview Fusion of Vision-Language Embeddings
CAMFusion fuses multiview 2D vision-language embeddings via cross-attention and multiview consistency self-supervision to produce better 3D semantic and instance representations, outperforming averaging and reaching SOTA on benchmarks including zero-shot out-of-domain cases.
-
ArtifactWorld: Scaling 3D Gaussian Splatting Artifact Restoration via Video Generation Models
ArtifactWorld restores artifacts in 3D Gaussian Splatting by training a video diffusion backbone on 107.5K paired clips with an isomorphic predictor for artifact heatmaps and an Artifact-Aware Triplet Fusion mechanism to achieve better sparse-view novel synthesis.
-
VVGT: Visual Volume-Grounded Transformer
VVGT is a dual-transformer network with Volume Geometry Forcing that maps volumetric data to 3D Gaussian primitives for accurate ray-based rendering without per-scene optimization.
-
Learning Long-term Motion Embeddings for Efficient Kinematics Generation
A 64x temporally compressed motion embedding learned from trackers enables efficient conditional flow-matching generation of long-term motions that outperform video models and task-specific methods.
-
Continuous Adversarial Flow Models
Continuous adversarial flow models replace MSE in flow matching with adversarial training via a discriminator, improving guidance-free FID on ImageNet from 8.26 to 3.63 for SiT and similar gains for JiT and text-to-image benchmarks.
-
DexWorldModel: Causal Latent World Modeling towards Automated Learning of Embodied Tasks
CLWM with DINOv3 targets, O(1) TTT memory, SAI latency masking, and EmbodiChain training achieves SOTA dual-arm simulation performance and zero-shot sim-to-real transfer that beats real-data finetuned baselines.
-
Self-supervised Pretraining of Cell Segmentation Models
DINOCell achieves a SEG score of 0.784 on LIVECell by self-supervised domain adaptation of DINOv2, improving 10.42% over SAM-based models and showing strong zero-shot transfer.
-
Enhancing Fine-Grained Spatial Grounding in 3D CT Report Generation via Discriminative Guidance
DCP-PD improves macro F1 scores on CT report generation benchmarks and introduces a hierarchical location-aware evaluation protocol that reveals ongoing challenges in pathology spatial grounding.
-
Adapting 2D Multi-Modal Large Language Model for 3D CT Image Analysis
Transferring a 2D MLLM to 3D CT inputs via parameter reuse, a Text-Guided Hierarchical MoE framework, and two-stage training yields better performance than prior 3D medical MLLMs on medical report generation and visual question answering.
-
ACCIDENT: A Benchmark Dataset for Vehicle Accident Detection from Traffic Surveillance Videos
ACCIDENT is a new benchmark with 2,027 real and 2,211 synthetic annotated video clips for temporal localization, spatial localization, and collision type classification of vehicle accidents in CCTV footage.
-
V-CAGE: Vision-Closed-Loop Agentic Generation Engine for Robotic Manipulation
V-CAGE automates the creation of scalable, high-quality robotic manipulation datasets through context-aware scene construction, closed-loop visual verification, and perceptually-driven compression.
-
Meta-learning In-Context Enables Training-Free Cross Subject Brain Decoding
A meta-optimized in-context learning approach enables training-free cross-subject semantic visual decoding from fMRI by inferring individual neural encoding patterns via hierarchical inference on a few examples.
-
Self-Improving 4D Perception via Self-Distillation
SelfEvo enables pretrained 4D perception models to self-improve on unlabeled videos via self-distillation, delivering up to 36.5% relative gains in video depth estimation and 20.1% in camera estimation across eight benchmarks.
-
TC-AE: Unlocking Token Capacity for Deep Compression Autoencoders
TC-AE improves reconstruction and generative performance in deep compression by decomposing token-to-latent compression into two stages and using joint self-supervised training.
-
GIRL: Generative Imagination Reinforcement Learning via Information-Theoretic Hallucination Control
GIRL reduces latent rollout drift by 38-61% versus DreamerV3 in MBRL by grounding transitions with DINOv2 embeddings and using an information-theoretic adaptive bottleneck, yielding better long-horizon returns on control benchmarks.
-
Data Warmup: Complexity-Aware Curricula for Efficient Diffusion Training
Data Warmup accelerates diffusion training on ImageNet by scheduling images from low to high complexity via a foreground-based metric and temperature-controlled sampler, improving FID and IS scores faster than uniform sampling.
-
Telescope: Learnable Hyperbolic Foveation for Ultra-Long-Range Object Detection
Telescope uses learnable hyperbolic foveation to deliver a 76% relative mAP gain (0.185 to 0.326) for objects beyond 250 meters while keeping overhead low.
-
RoboPlayground: Democratizing Robotic Evaluation through Structured Physical Domains
RoboPlayground reframes robotic manipulation evaluation as a language-driven process over structured physical domains, letting users author varied yet reproducible tasks that reveal policy generalization failures.
-
Part-Level 3D Gaussian Vehicle Generation with Joint and Hinge Axis Estimation
A new framework generates part-level animatable 3D Gaussian vehicles from images by adding modules for exclusive part ownership and kinematic joint/axis prediction.
-
Uncertainty-Aware Foundation Models for Clinical Data
The work introduces uncertainty-aware foundation models for clinical data by learning set-valued patient representations that enforce consistency across partial observations and integrate multimodal self-supervised objectives.
-
Learning Robust Visual Features in Computed Tomography Enables Efficient Transfer Learning for Clinical Tasks
VoxelFM learns robust 3D CT visual features via DINO self-distillation that transfer effectively to seven clinical task categories using frozen backbones and lightweight heads, outperforming prior CT foundation models even on report generation.
-
Task-Guided Multi-Annotation Triplet Learning for Remote Sensing Representations
Task-guided triplet selection via mutual information improves multi-annotation representation learning over static weighting on an aerial wildlife dataset.
-
Inference-Path Optimization via Circuit Duplication in Frozen Visual Transformers for Marine Species Classification
Circuit duplication on frozen DINOv3 embeddings raises macro F1 to 0.875 on AQUA20, within 1.4 points of supervised ConvNeXt, with class-specific circuits helping 75% of species.
-
Automated Segmentation and Tracking of Group Housed Pigs Using Foundation Models
Foundation models plus modular tracking achieve stable long-term pig segmentation and identity maintenance on farm videos with 0.99 MOTA and no switches.
-
CoME-VL: Scaling Complementary Multi-Encoder Vision-Language Learning
CoME-VL fuses contrastive and self-supervised vision encoders via entropy-guided multi-layer aggregation and RoPE cross-attention to improve vision-language model performance on benchmarks.
-
The Compression Gap: Why Discrete Tokenization Limits Vision-Language-Action Model Scaling
Discrete action tokenization in VLA models creates an information bottleneck that prevents vision encoder scaling from improving performance, unlike continuous policies, as validated on the LIBERO benchmark.
-
Gram-MMD: A Texture-Aware Metric for Image Realism Assessment
Gram-MMD is a texture-aware realism metric that computes MMD on upper-triangular Gram matrices from backbone activations, providing complementary information to semantic distributional metrics.
-
Rapidly deploying on-device eye tracking by distilling visual foundation models
DistillGaze reduces median gaze error by 58.62% on a 2000+ participant dataset by distilling foundation models into a 256K-parameter on-device model using synthetic labeled data and unlabeled real data.
-
UniRecGen: Unifying Multi-View 3D Reconstruction and Generation
UniRecGen unifies reconstruction and generation via shared canonical space and disentangled cooperative learning to produce complete, consistent 3D models from sparse views.
-
Chat-Scene++: Exploiting Context-Rich Object Identification for 3D LLM
Chat-Scene++ improves 3D scene understanding in multimodal LLMs by representing scenes as context-rich object sequences with identifier tokens and grounded chain-of-thought reasoning, reaching state-of-the-art on five benchmarks using pre-trained encoders.
-
HD-VGGT: High-Resolution Visual Geometry Transformer
HD-VGGT achieves state-of-the-art high-resolution 3D reconstruction from image collections via a dual-branch architecture that predicts coarse geometry at low resolution and refines details at high resolution while modulating unreliable features.
-
MPDiT: Multi-Patch Global-to-Local Transformer Architecture For Efficient Flow Matching and Diffusion Model
MPDiT uses a hierarchical multi-patch design in transformers to lower computation in diffusion models by handling coarse global features first then fine local details, plus faster-converging embeddings.
-
Emergent Neural Automaton Policies: Learning Symbolic Structure from Visuomotor Trajectories
ENAP extracts an emergent Mealy automaton from visuomotor trajectories to act as a high-level planner for a low-level residual policy, yielding up to 27% higher success than end-to-end VLA policies in low-data regimes.
-
Generative Event Pretraining with Foundation Model Alignment
GEP transfers semantic knowledge from image foundation models to event data via alignment and generative pretraining on mixed sequences to create transferable event-based visual models.
-
VISER: Visually-Informed System for Enhanced Robustness in Open-Set Iris Presentation Attack Detection
Denoised eye tracking heatmaps yield the largest generalization gain in open-set iris presentation attack detection, lowering APCER at 1% BPCER compared to cross-entropy training.
-
Continual Few-shot Adaptation for Synthetic Fingerprint Detection
A continual few-shot adaptation method combining binary cross-entropy and supervised contrastive losses with replay achieves a good trade-off between fast adaptation to unseen synthetic fingerprint styles and retention of known styles.
-
Causal Attribution via Activation Patching
CAAP produces patch attributions in ViTs by direct activation patching on intermediate layers to measure causal contribution to the target class score.
-
Diffusion Models Memorize in Training -- and Generalize in Inference
Diffusion models overfit denoising loss at intermediate noise but generalize in inference as model error smooths the flow field and sampling paths avoid memorized noisy training data.
-
RPG-SAM: Reliability-Weighted Prototypes and Geometric Adaptive Threshold Selection for Training-Free One-Shot Polyp Segmentation
RPG-SAM improves one-shot polyp segmentation by weighting high-fidelity support features and dynamically adjusting thresholds via morphological consensus, yielding 5.56% mIoU gain on Kvasir.
-
From Local Matches to Global Masks: Template-Guided Instance Detection and Segmentation in Open-World Scenes
L2G-Det detects and segments novel object instances in open scenes by using local template patch matches to generate points that prompt an augmented SAM for global masks.
-
Robometer: Scaling General-Purpose Robotic Reward Models via Trajectory Comparisons
Robometer combines intra-trajectory progress supervision with inter-trajectory preference supervision on a 1M-trajectory dataset to learn more generalizable robotic reward functions than prior methods.
-
InCoM: Intent-Driven Perception and Structured Coordination for Mobile Manipulation
InCoM achieves 23-28% higher success rates in mobile manipulation tasks by inferring motion intent for adaptive perception and decoupling base-arm action generation.
-
Drift Localization using Conformal Predictions
Conformal predictions enable drift localization by identifying affected samples, outperforming local testing on image datasets.
-
SurgMotion: A Video-Native Foundation Model for Universal Understanding of Surgical Videos
SurgMotion outperforms prior methods on 17 surgical video benchmarks by shifting pretraining to latent motion prediction with motion-guided masking, affinity distillation, and diversity regularization on a 15M-sample dataset.
-
Vision-aligned Latent Reasoning for Multi-modal Large Language Model
VaLR generates vision-aligned latent tokens before each reasoning step to preserve perceptual cues, improving VSI-Bench accuracy from 33.0% to 52.9%.