For stationary image datasets and standard augmentations, the optimal contrastive representation is partial whitening of DFT power, implemented by a CNN with sinusoidal first-layer filters and a waterfilling weight allocation.
hub Mixed citations
Journal of Foo , volume = 13, number = 1, pages =
Mixed citation behavior. Most common role is background (55%).
29
Pith papers citing it
Background
55% of classified citations
hub tools
citation-role summary
background 7
method 2
baseline 1
other 1
citation-polarity summary
representative citing papers
GeoMix achieves new state-of-the-art results in descriptor-free 2D-3D matching by adding directional embeddings, learnable global context nodes, and multi-detector training, cutting rotation and translation errors by up to 90% on standard benchmarks.
SymTrack is the first systematic detection-free framework for scene text tracking that constructs benchmarks from video text spotting datasets and reports up to 11.97% AUC gains over prior trackers.
Single-shot HDR is achieved by conditioning a video diffusion model on an LDR input to generate an exposure bracket and fusing the bracket with per-pixel weights from a lightweight UNet.
HairGPT reframes 3D hairstyle synthesis as dual-decoupled autoregressive strand sequence modeling with geometric tokenization for semantic control and rare style generation.
A text-supervised global layout embedding augments local patch representations in late-interaction VDR, yielding +2.4 nDCG@5 and +2.3 MAP@5 gains over ColPali/ColQwen baselines on ViDoRe-v2.
Privatar uses horizontal frequency partitioning and distribution-aware minimal perturbation to enable private offloading of VR avatar reconstruction, supporting 2.37x more users with modest overhead.
Q-Align trains LMMs on discrete text-defined levels for visual scoring, achieving SOTA on IQA, IAA, and VQA while unifying the tasks in OneAlign.
Sequential testing with tailored stopping rules lets model evaluation halt early once statistical needs (CI width, significance, equivalence) are met, saving up to 80% compute on VLM leaderboards.
InduceKV is a retrieval-based continual adaptation method that uses bilevel selection to build a compact set of inducing KV memories for fixed-footprint updates to multimodal LLMs.
The paper defines translation-invariant calibrators and class-conditional affine calibrators under argmax- or order-preservation constraints for semantic segmentation, showing improved calibration without segmentation degradation on benchmarks.
VideoSearch-R1 achieves SOTA on VCMR across three datasets via iterative retrieval, latent-space soft query refinement, and GRPO training.
A hardware prototype performs gaze estimation by optically encoding task-relevant features with a microlens array and mask, captured on a 4x4 phototransistor array and decoded by a small neural network, reaching 3.4 ms latency with competitive accuracy.
Pilot study uses pretrained video encoder features from lung ultrasound to predict 30-day CHF readmission, finding lower-lung views and temporal differences most informative with top MLP F1 of 0.80.
DocAtlas introduces model-free rendering pipelines to create DocTag-annotated datasets across 82 languages and shows DPO adaptation improves multilingual performance without base-language degradation.
ECTraj delivers competitive Argoverse-2 multi-agent trajectory metrics via single-step consistency models that are ~10× faster than multi-step diffusion baselines.
ProCompNav builds a candidate pool from ambiguous queries then uses pool-splitting binary questions for disambiguation, improving success rate and shortening responses on CoIN-Bench and TextNav.
LIVEditor-14B applies a new sparse attention method (ISA) that prunes context and uses query-sharpness routing to cut attention latency ~60% with no loss in editing quality on standard benchmarks.
Privatar partitions VR avatar reconstruction via frequency-domain decomposition, keeping sensitive components local and offloading the rest with distribution-aware minimal perturbation noise, achieving 2.37x throughput with provable privacy.
Video Prediction Policy conditions robot action learning on future-frame predictions inside fine-tuned video diffusion models, yielding 18.6% relative gains on Calvin ABC-D and 31.6% higher real-world success rates.
Biased noise sampling for rectified flows combined with a bidirectional text-image transformer architecture yields state-of-the-art high-resolution text-to-image results that scale predictably with model size.
A self-supervised approach uses consistent spatial relationships of anatomical structures across patients to improve 3D multi-modal medical image representations, yielding modest gains on segmentation and classification tasks.
A UAV-to-3DGS-to-MPM pipeline reconstructs real landslide sites with photorealistic visuals and runs physics-based simulations, validated on a Hong Kong event.
ActDiff-VC partitions video into segments, transmits adaptive keyframes and budget-aware point trajectories, and reconstructs frames via conditional diffusion, reporting up to 64.6% bitrate reduction at matched NIQE on UVG and MCL-JCV.
citing papers explorer
-
A Theory of Contrastive Learning with Natural Images
For stationary image datasets and standard augmentations, the optimal contrastive representation is partial whitening of DFT power, implemented by a CNN with sinusoidal first-layer filters and a waterfilling weight allocation.
-
GeoMix: Descriptor-Free Visual Localization via Global Context and Multi-Detector Training
GeoMix achieves new state-of-the-art results in descriptor-free 2D-3D matching by adding directional embeddings, learnable global context nodes, and multi-detector training, cutting rotation and translation errors by up to 90% on standard benchmarks.
-
Beyond Detection: A Structure-Aware Framework for Scene Text Tracking
SymTrack is the first systematic detection-free framework for scene text tracking that constructs benchmarks from video text spotting datasets and reports up to 11.97% AUC gains over prior trackers.
-
Single-Shot HDR Recovery via a Video Diffusion Prior
Single-shot HDR is achieved by conditioning a video diffusion model on an LDR input to generate an exposure bracket and fusing the bracket with per-pixel weights from a lightweight UNet.
-
HairGPT: Strand-as-Language Autoregressive Modeling for Realistic 3D Hairstyle Synthesis
HairGPT reframes 3D hairstyle synthesis as dual-decoupled autoregressive strand sequence modeling with geometric tokenization for semantic control and rare style generation.
-
Beyond Bag-of-Patches: Learning Global Layout via Textual Supervision for Late-Interaction Visual Document Retrieval
A text-supervised global layout embedding augments local patch representations in late-interaction VDR, yielding +2.4 nDCG@5 and +2.3 MAP@5 gains over ColPali/ColQwen baselines on ViDoRe-v2.
-
Privatar: Scalable Privacy-preserving Multi-user VR via Secure Offloading
Privatar uses horizontal frequency partitioning and distribution-aware minimal perturbation to enable private offloading of VR avatar reconstruction, supporting 2.37x more users with modest overhead.
-
Q-Align: Teaching LMMs for Visual Scoring via Discrete Text-Defined Levels
Q-Align trains LMMs on discrete text-defined levels for visual scoring, achieving SOTA on IQA, IAA, and VQA while unifying the tasks in OneAlign.
-
Stop Guessing When to Stop Testing: Efficient Model Evaluation with Just Enough Data
Sequential testing with tailored stopping rules lets model evaluation halt early once statistical needs (CI width, significance, equivalence) are met, saving up to 80% compute on VLM leaderboards.
-
InduceKV: Fixed-Footprint Continual Adaptation of Multimodal LLMs via Inducing KV Memories
InduceKV is a retrieval-based continual adaptation method that uses bilevel selection to build a compact set of inducing KV memories for fixed-footprint updates to multimodal LLMs.
-
Rethinking Post-Hoc Calibration in Semantic Segmentation
The paper defines translation-invariant calibrators and class-conditional affine calibrators under argmax- or order-preservation constraints for semantic segmentation, showing improved calibration without segmentation degradation on benchmarks.
-
VideoSearch-R1: Iterative Video Retrieval and Reasoning via Soft Query Refinement
VideoSearch-R1 achieves SOTA on VCMR across three datasets via iterative retrieval, latent-space soft query refinement, and GRPO training.
-
Low Latency Gaze Tracking via Latent Optical Sensing
A hardware prototype performs gaze estimation by optically encoding task-relevant features with a microlens array and mask, captured on a 4x4 phototransistor array and decoded by a small neural network, reaching 3.4 ms latency with competitive accuracy.
-
Prognostic Value of Lung Ultrasound Biomarkers for Readmission Risk in Congestive Heart Failure: A Pilot Data-Driven Analysis
Pilot study uses pretrained video encoder features from lung ultrasound to predict 30-day CHF readmission, finding lower-lung views and temporal differences most informative with top MLP F1 of 0.80.
-
DocAtlas: Multilingual Document Understanding Across 80+ Languages
DocAtlas introduces model-free rendering pipelines to create DocTag-annotated datasets across 82 languages and shows DPO adaptation improves multilingual performance without base-language degradation.
-
ECTraj: Enhanced Consistency Training for Multi-Agent Trajectory Prediction
ECTraj delivers competitive Argoverse-2 multi-agent trajectory metrics via single-step consistency models that are ~10× faster than multi-step diffusion baselines.
-
ProCompNav: Proactive Instance Navigation with Comparative Judgment for Ambiguous User Queries
ProCompNav builds a candidate pool from ambiguous queries then uses pool-splitting binary questions for disambiguation, improving success rate and shortening responses on CoIN-Bench and TextNav.
-
LIVEditor-14B: Lightning Unified Video Editing via In-Context Sparse Attention
LIVEditor-14B applies a new sparse attention method (ISA) that prunes context and uses query-sharpness routing to cut attention latency ~60% with no loss in editing quality on standard benchmarks.
-
Dual-Anchoring: Addressing State Drift in Vision-Language Navigation
Privatar partitions VR avatar reconstruction via frequency-domain decomposition, keeping sensitive components local and offloading the rest with distribution-aware minimal perturbation noise, achieving 2.37x throughput with provable privacy.
-
Video Prediction Policy: A Generalist Robot Policy with Predictive Visual Representations
Video Prediction Policy conditions robot action learning on future-frame predictions inside fine-tuned video diffusion models, yielding 18.6% relative gains on Calvin ABC-D and 31.6% higher real-world success rates.
-
Scaling Rectified Flow Transformers for High-Resolution Image Synthesis
Biased noise sampling for rectified flows combined with a bidirectional text-image transformer architecture yields state-of-the-art high-resolution text-to-image results that scale predictably with model size.
-
Beyond Instance-Level Self-Supervision in 3D Multi-Modal Medical Imaging
A self-supervised approach uses consistent spatial relationships of anatomical structures across patients to improve 3D multi-modal medical image representations, yielding modest gains on segmentation and classification tasks.
-
UAV-Assisted Scan-to-Simulation for Landslides Using Physics-Informed Gaussian Splatting
A UAV-to-3DGS-to-MPM pipeline reconstructs real landslide sites with photorealistic visuals and runs physics-based simulations, validated on a Hong Kong event.
-
Active Sampling for Ultra-Low-Bit-Rate Video Compression via Conditional Controlled Diffusion
ActDiff-VC partitions video into segments, transmits adaptive keyframes and budget-aware point trajectories, and reconstructs frames via conditional diffusion, reporting up to 64.6% bitrate reduction at matched NIQE on UVG and MCL-JCV.
-
Intrinsic Gradient Suppression for Label-Noise Prompt Tuning in Vision-Language Models
Double-Softmax Prompt Tuning uses sequential softmax normalization to create self-adaptive gradient saturation that filters noisy samples while preserving useful updates in CLIP prompt tuning.
-
LTM: Large-scale Terrain Model for Wildfire-prone Landscapes
A ray-tracing pipeline aligns ground-level image pixels to outdated DEM rasters for real-time 3D terrain reconstruction in wildfire zones, validated primarily through a custom simulator.
-
ELEVATE: Designing Human-Centered GenAI Virtual Tutors for Scalable and Inclusive Education
ELEVATE is a framework and prototype for deploying LLM-powered 3D avatar tutors locally on consumer hardware with a three-stratum design separating interaction, execution, and governance layers.
-
Bad Seeing or Bad Thinking? Rewarding Perception for Multimodal Reasoning
Proposes Modality-Aware Credit Assignment (MoCA) with blindfolded-reasoning proxy to reward perception fidelity separately from reasoning in VLMs.
-
Low-Cost Neural Radiance Fields
Comparative study of DS-NeRF, TensoRF, and HashNeRF with depth-supervision and architectural variants finds no conclusive outperformance under equal training time but identifies which design choices transfer to low-data, low-compute regimes.