Computer Vision - Weekly Roundup

Diffusion models can learn lattice field theories without ever seeing real data, matching HMC performance.

A single dimensionless motion prior generalizes across vastly different object families and data modalities, outperforming specialized methods without retraining.

A single latent cache can unify diverse visual representations and spatiotemporal queries for efficient, consistent dynamic world modeling across multiple video generation tasks.

Generate graphs that *exactly* realize a target persistence diagram, moving beyond approximate topological matching.

A single MLLM can now perform temporal grounding and video QA with significant inference speedups by treating them as structured decision tasks.

Diffusion models can learn lattice field theories without ever seeing real data, matching HMC performance.

A single dimensionless motion prior generalizes across vastly different object families and data modalities, outperforming specialized methods without retraining.

A single latent cache can unify diverse visual representations and spatiotemporal queries for efficient, consistent dynamic world modeling across multiple video generation tasks.

Generate graphs that *exactly* realize a target persistence diagram, moving beyond approximate topological matching.

A single MLLM can now perform temporal grounding and video QA with significant inference speedups by treating them as structured decision tasks.

LLMs can now actively seek and ground verifiable spatio-temporal evidence, including object bounding boxes, to improve video reasoning accuracy by over 12%.

Generative 3D reconstruction now scales to entire indoor scenes, producing coherent, unified meshes with state-of-the-art fidelity from unposed images.

An agent can perform city-scale 3D object grounding by actively exploring a multi-view environment, bypassing the need for precomputed proposals and achieving state-of-the-art results.

Aesthetic code generation is unlocked by using a VLM judge to translate visual design principles from a reference image into actionable rewards for code models.

Humanoid robots can now learn complex tasks from egocentric, multimodal human demonstrations thanks to a wearable observatory and a novel motion reconstruction pipeline that achieves state-of-the-art accuracy.

Image tokens can be compressed by 94% by prioritizing semantic detail over spatial uniformity, unlocking efficient vision-language models.

Fine-grained, foundation-model-derived semantic representations unlock state-of-the-art whole-body PET/CT registration without task-specific fine-tuning.

VLMs can be taught to *use* latent geometry by training them to read continuous geometric tokens, rather than just reconstructing geometry from RGB.

Training-free conditional sampling for Schrödinger Bridges is now possible, achieving state-of-the-art results on image generation and inverse problems by twisting SB's inherent endpoint prediction capabilities.

Mixing the right local appearances, not just blending pixels, unlocks significantly higher quality texture interpolations.

A single dataset now unifies 279,166 multi-sensor, multi-temporal Earth observation tiles, unlocking efficient pretraining for generalist foundation models.

Physical controls like damping and size can precisely guide audio foley generation, achieving near-perfect correlation with real-world sound targets without fine-tuning.

Segmentation models implicitly learn powerful object-centric representations that outperform encoder-based feature extraction for medical image prediction, even without explicit downstream supervision.

Robust generative modeling is achieved by replacing squared error with geometric median regression in Flow Matching, outperforming standard methods under severe data contamination.

Physically informed GNNs can accurately forecast advection-diffusion dynamics on irregular grids, outperforming existing methods and enhancing solar power predictions.

Data replay methods can recover up to 88% of performance in task-incremental radar waveform classification, with synergistic combinations of replay mechanisms boosting accuracy by over 5%.

Two residual streams in diffusion language models halve the performance gains of adding a second stream by enabling distinct read and write routing patterns.

Most industrial anomaly detectors collapse under simple brightness or zoom changes, losing up to 34 AUROC points, but a new approach using synthetic anomalies maintains over 92% AUROC.

Generative image and video compression achieves significant memory and compute reduction by replacing dense Gaussian codebooks with implicit randomized Hadamard dictionaries, enabling real-time streaming video compression on a single GPU.

Few-step diffusion models can achieve state-of-the-art 4-NFE generation without auxiliary score models or adversarial training, matching or exceeding multi-step teachers.

Ultralight action heads achieve state-of-the-art embodied task performance by learning to selectively retrieve visual information via shared latent bridge queries.

LLMs can now understand and execute complex, semantic image edits from sparse drag inputs by inferring intended object-level transformations.

Image reconstruction and generation quality rivaling dense tokenizers are achieved with ~130 tokens by adaptively refining global 1D representations with sparse 2D tokens.

Heterogeneous diffusion models can be seamlessly stitched together, unlocking new capabilities by sharing a common latent language.

Generative representations need more than just semantic alignment; they require active exploitation, as shown by Representation Guidance outperforming larger models on ImageNet with a novel flow-matching target modification.

Beyond transcribing words, a novel convolutional-Transformer model now timestamps non-verbal vocalizations and background sounds, enriching ASR with a richer auditory context.

Abstracting human demonstrations into scene graphs unlocks zero-shot generalization for dexterous manipulation policies, achieving 71% better performance on unseen tasks than imitation-heavy methods.

A single recurrent Transformer block can match the accuracy of deep vision encoders with 70% fewer parameters by dynamically programming FFN experts via a continuous depth coordinate.

A single visual world model can now execute complex procedural tasks end-to-end, adapting its generation to the actual outcome of each predicted action.

Explicit 3D geometry, rendered as text, unlocks superior spatial reasoning in RGB-only VLMs by guiding trajectory correction during distillation.

Fine-grained 3D object generation is now possible by directly controlling local geometry and appearance with text and shape primitives, bypassing the need for model retraining.

Direct 2D spatial composition with four binding types unlocks unprecedented controllability in poster generation, outperforming prompt-based methods.

A single regional weather model can now leverage multiple global forecast drivers without retraining, achieving kilometer-scale accuracy.

Ambiguous object selection in text-guided image editing is solved by directly annotating the canvas, achieving state-of-the-art performance in complex edits.

A reconstruction-free JEPA successfully disentangles invariant and variant latent factors, offering a path to robust world models without observational decoders.

City-scale audio-visual reasoning is currently dominated by expensive agentic pipelines, not standalone multimodal models.

A training-free change detection method achieves near-optimal performance by integrating diffusion models to precisely estimate Hyvärinen scores, slashing detection delays by over 90% in simulations.

End-to-end diffusion models can achieve state-of-the-art precipitation nowcasting with over 17x faster inference by forecasting directly in radar space.

Iterative refinement in data space, not flow space, unlocks fast, high-fidelity generative sampling without manual timestep tuning.

Contextual contamination in few-shot vision-language learning is mitigated by a novel semantic purification and reinforcement-learning gating mechanism, unlocking superior performance.

A frozen video tokenizer's latent can be factored into controllable and reward-relevant components, unlocking significant downstream planning gains.

Egocentric video generation can be achieved by a learned transformer that probabilistically synthesizes novel views, outperforming explicit depth-based reconstruction pipelines.

Generated robot futures can actively harm performance if temporally misaligned, but a new control-based conditioning method recovers substantial success by learning to trust and align them.

Casual images are now enough to generate high-fidelity, scene-aligned 3D objects with accurate pose, even from sparse observations.

Out-of-distribution data can dramatically boost discrete diffusion model performance under extreme data scarcity by strategically injecting it at low noise levels.

Local contact reasoning, not global scene modeling, unlocks 30%+ accuracy gains in rigid-body physics simulation.

Generative models can now capture the full spectrum of solutions in high-dimensional physical systems with symmetry-breaking bifurcations, a feat previously intractable.

Static 3D worlds now possess vivid, looping motion, seamlessly transitioning from any viewpoint without explicit motion capture.

Satellite imagery unlocks fine-grained socioeconomic predictions with calibrated uncertainty, even without direct ground truth.

Diffusion models bypass Langevin's condition-number bottleneck by leveraging time-dependent score trajectories, achieving sampling error independent of target covariance ill-conditioning.

Deterministic anomaly detection with RIFT precisely matches Isolation Forest's theoretical limit in 1D and generalizes robustly to higher dimensions, outperforming IF's variance in ensemble settings.

Decoding schedules for diffusion and autoregressive models can be ranked by a kernel of pairwise dependence estimated from pretrained weights, predicting performance across modalities before generation.

Most 3D foundation models struggle to accurately estimate absolute metric distances, a fundamental capability overlooked by existing benchmarks.

Frontier MLLMs exhibit strong local perception but falter on long-horizon audio-visual reasoning, with a single benchmark exposing specific failure modes like identity drift and cross-modal misalignment.

Identity-preserving dense correspondence is now achievable across arbitrary image transformations, even those breaking physical continuity, by leveraging foundation models and multi-source supervision.

A unified multimodal agent, OneSearch-VL, achieves state-of-the-art performance on diverse visual research tasks by explicitly modeling and preserving interdependencies within a Visually Grounded Evidence Graph.

Complex-valued fusion of audio-visual embeddings unlocks significant gains in open-vocabulary event localization, surpassing prior methods by up to 6.6% Event-F1.

Realistic human-object interactions in complex 3D scenes are now achievable by decoupling scene feasibility from motion synthesis, leveraging separate supervision signals.

Humanoid robots can now catch thrown objects with 65% success from 2.5m using only egocentric RGB-D, a leap in reactive whole-body control.

DiffusionART achieves state-of-the-art RL fine-tuning for diffusion models, reaching peak performance with over 12x fewer GPU hours and substantially higher rewards than existing methods.

Achieve state-of-the-art image generation quality at a fraction of the compute cost by dynamically adapting tokenization to visual complexity.

Even state-of-the-art MLLMs struggle to grasp the nuanced data interpretation and narrative coherence required for data videos, with significant performance gaps in causal reasoning and narrative structure.

Decouple complex audio scenes into individual, controllable sound sources directly from a single generative model.

Static video regions can be denoised 50% faster without sacrificing quality by dynamically skipping computation on easy patches.

An 8x token reduction and 11.1x latency improvement in video generation is achieved by training autoencoders that are generation-aware, not just reconstruction-aware.

Current image generation models struggle with rendering dense, multi-region text, with fidelity dropping by over 14 points when clarity is prioritized.

A quadtree visual tokenizer slashes token count for autoregressive image generation by adaptively focusing representational power on intricate image regions.

Stored patch vectors from visual document retrieval systems can be inverted to reconstruct original pages, revealing sensitive information and enabling source page identification with high accuracy.

Decoupling denoising and context writing parameters in autoregressive video models unlocks unprecedented long-horizon generation capabilities, extending continuous output to 24 hours from just 5-second training rollouts.

Global point set alignment is solved exactly by finding the farthest vertex of a novel "permutation polygon," resolving a long-standing open problem in assignment geometry.

Generating 3D scenes where objects not only look plausible but also physically fit together, even when interactions are hidden.

Autoregressive video pretraining unlocks the full potential of long contexts for real-time robot control, boosting success rates by 15% and enabling complex manipulation previously impossible.

Evolutionary search discovers a 26x smaller, yet more accurate, model for satellite-based chlorophyll prediction than expert-designed alternatives.

A 300K-sample dataset and diagnostic benchmark unlock the challenging task of generating product-preserving, narrative-driven advertisement videos.

Fresh, shallow network activations can correct diffusion model feature forecasts, unlocking significant speedups and improved fidelity without any parameter updates.

Probabilistic forecasting models can now achieve 65% CRPS reduction with drastically fewer samples by directly learning the distribution shift between successive predictions.

Generative AI now predicts complex 3D urban wind and temperature fields with LES-level accuracy in seconds, unlocking rapid, turbulence-aware urban design.

Fingertip touch dynamics, not just vision, unlock a 10.5% boost in dexterous robotic grasp success.

Transformers learn translation, rotation, and scale invariance through distinct, identifiable mechanisms, paving the way for structured equivariant function learning.

Inertial activity recognition models can now learn rich, global motion structure from unlabeled data by discovering and grouping motion primitives, achieving state-of-the-art performance with orders of magnitude less data.

Diffusion models can achieve arbitrarily low error amplification at low noise levels, decoupling sampler performance from worst-case Lipschitz bounds.

Spatially dependent data breaks standard conformal prediction; this method achieves exact coverage and oracle efficiency by sequentially whitening residuals, even with nonrandom sampling.

Kinetic Langevin diffusions slash posterior sampling time for imaging inverse problems by 10x while maintaining reconstruction quality, even with complex diffusion model priors.

Embodied agents can now remember objects they never touched by tracking their resting locations and inferring moves from repeated evidence, boosting downstream task accuracy by over 10%.

Interactive video generation systems can now robustly handle latent dependencies by transforming unstructured inputs into executable constraints, achieving a 10.3-point average improvement.

A single generative model recovers temporally consistent 3D hand motion from video by jointly learning 2D-3D pose dynamics, bypassing unreliable per-frame estimates.

Sparse video reasoning can outperform dense observation models by learning to extract crucial temporal evidence from limited frames, even without explicit VQA annotations.

Interactive video generation can be sped up by 1.4x by intelligently caching transformer residuals based on detected control transitions, without sacrificing quality.

A single image can now yield complete, high-fidelity 3D scenes, indoors and out, by adapting object-centric priors with adaptive chunking and explicit 2D-3D correspondence.

Inserted objects can now be made "alive" and interact realistically within videos, solving a long-standing challenge in video editing.

Diffusion models reveal feature generation order quantitatively, exposing architectural differences and suggesting ordered training benefits.

Deterministic dermoscopic preprocessing actually degrades performance in leakage-controlled evaluations, suggesting current automated analysis might be over-engineering for minimal gains.

Rare event probabilities in diffusion models can now be estimated orders of magnitude faster, unlocking their use in high-stakes scientific simulation.

The geometric intuition behind SVD is revealed as the direct algorithmic engine driving PCA, kernel methods, and PageRank, unifying their proofs and implementations.

Edge CNN adaptation can now achieve near-zero activation memory overhead by decoupling backward computations from full-width layer inputs.