ICCV 2025 accepted papers on Label-efficient and robust learning
213 of the 2,701 papers accepted at ICCV 2025 (7.9%) are on Label-efficient and robust learning, compared with 14.8% at ICCV 2023. They fall into 9 topics.
See ICCV 2025 in the interactive map
Submitting to ICCV? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Recognition and long-tail · 75 papers
- End-to-End Driving with Online Trajectory Evaluation via BEV World Model (118 citations)
- World4Drive: End-to-End Autonomous Driving via Intention-aware Physical Latent World Model (60 citations)
- FaceXFormer: A Unified Transformer for Facial Analysis (58 citations)
- Hydra-NeXt: Robust Closed-Loop Driving with Open-Loop Training (37 citations)
- HiP-AD: Hierarchical and Multi-Granularity Planning with Deformable Attention for Autonomous Driving in a Single Decoder (28 citations)
- Where, What, Why: Towards Explainable Driver Attention Prediction (21 citations)
- Feature Coding in the Era of Large Models: Dataset, Test Conditions, and Benchmark (16 citations)
- LVFace: Progressive Cluster Optimization for Large Vision Models in Face Recognition (13 citations)
- Unraveling the Effects of Synthetic Data on End-to-End Autonomous Driving (13 citations)
- Weakly Supervised Visible-Infrared Person Re-Identification via Heterogeneous Expert Collaborative Consistency Learning (13 citations)
- OpenAnimals: Revisiting Person Re-Identification for Animals Towards Better Generalization (11 citations)
- On the Provable Importance of Gradients for Autonomous Language-Assisted Image Clustering (10 citations)
and 63 more in the interactive map.
Domain adaptation and generalization · 32 papers
- UMDATrack: Unified Multi-Domain Adaptive Tracking Under Adverse Weather Conditions (14 citations)
- Adapting In-Domain Few-Shot Segmentation to New Domains without Source Domain Retraining (10 citations)
- Stronger, Steadier & Superior: Geometric Consistency in Depth VFM Forges Domain Generalized Semantic Segmentation (9 citations)
- PHATNet: A Physics-guided Haze Transfer Network for Domain-adaptive Real-world Image Dehazing (8 citations)
- Adversarial Data Augmentation for Single Domain Generalization via Lyapunov Exponent-Guided Optimization (5 citations)
- Exploring Weather-aware Aggregation and Adaptation for Semantic Segmentation under Adverse Conditions (5 citations)
- AdaDCP: Learning an Adapter with Discrete Cosine Prior for Clear-to-Adverse Domain Generalization (4 citations)
- Toward Fair and Accurate Cross-Domain Medical Image Segmentation: A VLM-Driven Active Domain Adaptation Paradigm (4 citations)
- Exploiting Domain Properties in Language-Driven Domain Generalization for Semantic Segmentation (3 citations)
- Exploring Probabilistic Modeling Beyond Domain Generalization for Semantic Segmentation (3 citations)
- Group-wise Scaling and Orthogonal Decomposition for Domain-Invariant Feature Extraction in Face Anti-Spoofing (3 citations)
- M2EIT: Multi-Domain Mixture of Experts for Robust Neural Inertial Tracking (3 citations)
and 20 more in the interactive map.
OOD and anomaly detection · 29 papers
- SeaS: Few-shot Industrial Anomaly Image Generation with Separation and Sharing Fine-tuning (27 citations)
- G2SF: Geometry-Guided Score Fusion for Multimodal Industrial Anomaly Detection (14 citations)
- MultiADS: Defect-aware Supervision for Multi-type Anomaly Detection and Segmentation in Zero-Shot Learning (14 citations)
- RareCLIP: Rarity-aware Online Zero-shot Industrial Anomaly Detection (14 citations)
- Salvaging the Overlooked: Leveraging Class-Aware Contrastive Learning for Multi-Class Anomaly Detection (14 citations)
- Training-Free Industrial Defect Generation with Diffusion Models (12 citations)
- Wave-MambaAD: Wavelet-driven State Space Model for Multi-class Unsupervised Anomaly Detection (10 citations)
- DecAD: Decoupling Anomalies in Latent Space for Multi-Class Unsupervised Anomaly Detection (7 citations)
- Towards Real Unsupervised Anomaly Detection Via Confident Meta-Learning (7 citations)
- Enpowering Your Pansharpening Models with Generalizability: Unified Distribution is All You Need (6 citations)
- Secure On-Device Video OOD Detection Without Backpropagation (6 citations)
- Activation Subspaces for Out-of-Distribution Detection (5 citations)
and 17 more in the interactive map.
Knowledge distillation · 19 papers
- Frequency-Aligned Knowledge Distillation for Lightweight Spatiotemporal Forecasting (73 citations)
- Knowledge Distillation with Refined Logits (25 citations)
- ACAM-KD: Adaptive and Cooperative Attention Masking for Knowledge Distillation (22 citations)
- EA-KD: Entropy-based Adaptive Knowledge Distillation (16 citations)
- Local Dense Logit Relations for Enhanced Knowledge Distillation (15 citations)
- General Compression Framework for Efficient Transformer Object Tracking (12 citations)
- Cross-Architecture Distillation Made Simple with Redundancy Suppression (11 citations)
- VRM: Knowledge Distillation via Virtual Relation Matching (9 citations)
- Evidential Knowledge Distillation (8 citations)
- Knowledge Distillation for Learned Image Compression (8 citations)
- Perspective-Aware Teaching: Adapting Knowledge for Heterogeneous Distillation (4 citations)
- VehicleMAE: View-asymmetry Mutual Learning for Vehicle Re-identification Pre-training via Masked AutoEncoders (3 citations)
and 7 more in the interactive map.
Semi-supervised learning · 17 papers
- CoTracker3: Simpler and Better Point Tracking by Pseudo-Labelling Real Videos (216 citations)
- Semi-supervised Concept Bottleneck Models (20 citations)
- When Confidence Fails: Revisiting Pseudo-Label Selection in Semi-supervised Semantic Segmentation (13 citations)
- Pseudo-SD: Pseudo Controlled Stable Diffusion for Semi-Supervised and Cross-Domain Semantic Segmentation (8 citations)
- Adaptive Learning of High-Value Regions for Semi-Supervised Medical Image Segmentation (7 citations)
- S4M: Boosting Semi-Supervised Instance Segmentation with SAM (7 citations)
- ConformalSAM: Unlocking the Potential of Foundational Segmentation Models in Semi-Supervised Semantic Segmentation with Conformal Prediction (6 citations)
- Two Losses, One Goal: Balancing Conflict Gradients for Semi-supervised Semantic Segmentation (5 citations)
- Divide-and-Conquer for Enhancing Unlabeled Learning, Stability, and Plasticity in Semi-supervised Continual Learning (4 citations)
- Learnable Logit Adjustment for Imbalanced Semi-Supervised Learning under Class Distribution Mismatch (4 citations)
- Self-Reinforcing Prototype Evolution with Dual-Knowledge Cooperation for Semi-Supervised Lifelong Person Re-Identification (4 citations)
- CaliMatch: Adaptive Calibration for Improving Safe Semi-supervised Learning (2 citations)
and 5 more in the interactive map.
Self-supervised and contrastive · 15 papers
- An OpenMind for 3D Medical Vision Self-supervised Learning (26 citations)
- SignRep: Enhancing Self-Supervised Sign Representations (19 citations)
- Structure-aware Semantic Discrepancy and Consistency for 3D Medical Image Self-supervised Learning (9 citations)
- Beyond [cls]: Exploring the True Potential of Masked Image Modeling Representations (8 citations)
- Prototype-based Contrastive Learning with Stage-wise Progressive Augmentation for Self-Supervised Fine-Grained Learning (4 citations)
- CABLD: Contrast-Agnostic Brain Landmark Detection with Consistency-Based Regularization (3 citations)
- Coordinate-based Speed of Sound Recovery for Aberration-Corrected Photoacoustic Computed Tomography (3 citations)
- SAMora: Enhancing SAM through Hierarchical Self-Supervised Pre-Training for Medical Images (3 citations)
- AIM: Amending Inherent Interpretability via Self-Supervised Masking (2 citations)
- ASCENT: Annotation-free Self-supervised Contrastive Embeddings for 3D Neuron Tracking in Fluorescence Microscopy (2 citations)
- Dataset Ownership Verification for Pre-trained Masked Models (2 citations)
- GECO: Geometrically Consistent Embedding with Lightspeed Inference (2 citations)
and 3 more in the interactive map.
Test-time adaptation · 11 papers
- BATCLIP: Bimodal Online Test-Time Adaptation for CLIP (18 citations)
- Purge-Gate: Backpropagation-Free Test-Time Adaptation for Point Clouds Classification via Token purging (6 citations)
- ETA: Energy-based Test-time Adaptation for Depth Completion (5 citations)
- Hybrid-TTA: Continual Test-time Adaptation via Dynamic Domain Shift Detection (5 citations)
- Progressive Test Time Energy Adaptation for Medical Image Segmentation (4 citations)
- DC-TTA: Divide-and-Conquer Framework for Test-Time Adaptation of Interactive Segmentation (3 citations)
- Processing and acquisition traces in visual encoders: What does CLIP know about your camera? (3 citations)
- TopoTTA: Topology-Enhanced Test-Time Adaptation for Tubular Structure Segmentation (3 citations)
- FRET: Feature Redundancy Elimination for Test Time Adaptation (2 citations)
- Synchronizing Task Behavior: Aligning Multiple Tasks during Test-Time Training (1 citation)
- Test-time Adaptation for Foundation Medical Segmentation Model Without Parametric Updates (1 citation)
Zero-/few-shot and meta-learning · 10 papers
- TikZero: Zero-Shot Text-Guided Graphics Program Synthesis (15 citations)
- SVIP: Semantically Contextualized Visual Patches for Zero-Shot Learning (14 citations)
- Test-Time Prompt Tuning for Zero-Shot Depth Completion (6 citations)
- Is Meta-Learning Out? Rethinking Unsupervised Few-Shot Classification with Limited Entropy (5 citations)
- Hierarchical Material Recognition from Local Appearance (4 citations)
- On the Generalization of Representation Uncertainty in Earth Observation (4 citations)
- Language-Driven Multi-Label Zero-Shot Learning with Semantic Granularity (3 citations)
- A Conditional Probability Framework for Compositional Zero-shot Learning (2 citations)
- Rethinking Few Shot CLIP Benchmarks: A Critical Analysis in the Inductive Setting (2 citations)
- Zero-Shot Compositional Video Learning with Coding Rate Reduction (1 citation)
Noisy labels · 5 papers
- CA2C: A Prior-Knowledge-Free Approach for Robust Label Noise Learning via Asymmetric Co-learning and Co-training (7 citations)
- Joint Asymmetric Loss for Learning with Noisy Labels (4 citations)
- Meta-Learning Dynamic Center Distance: Hard Sample Mining for Learning with Noisy Labels (4 citations)
- Boosting Class Representation via Semantically Related Instances for Robust Long-Tailed Learning with Noisy Labels (3 citations)
- Robust Dataset Condensation using Supervised Contrastive Learning
