Visual Representation Learning

How far a pushed object slides depends on its mass and friction, which no single image reveals. Pretrained visual encoders are increasingly used as the perception front end of world models for manipulation, and their physical competence is assessed with perturbation benchmarks and linear probes, almost always in a clean, fixed-camera scene. We show that these assessments cannot distinguish an encoder that infers physics from one that does not. CALIPER (calibrate, then predict) is a direct test: an object of unknown mass and friction is struck twice at known speeds, a third strike is shown only up to the moment of contact, and a linear readout on frozen features must predict how far the object slides. Swapping in another object's calibration clips checks that the evidence is actually used. Across 2,000 simulated episodes and eight representations, from V-JEPA 2 to a randomly initialised ViT and raw pixels, calibration adds +0.50 R^2 and the swap removes it. Yet in the clean scene every representation lands within 0.02 R^2 of the ceiling set by true simulator state, because a fixed camera exposes the object's displacement directly in pixel coordinates. Resampling camera, lighting, and clutter for every clip spreads the same representations across 0.50 R^2; when the readout chooses a push speed for a goal distance, V-JEPA 2 misses by 4 mm and the random ViT by 20 mm, no better than ignoring the object. Linear probes track none of this: a change in frame aggregation moves a probe more than pretraining does, and erasing the probed mass direction from the same representation costs nothing in one scene and 0.35 R^2 in the other. Whether a benchmark can rank models is an empirical property, and we give three checks that establish it.

Vision-language pre-training and predictive world modeling provide robot policies with rich semantic and dynamic visual features, but their native action and visual-prediction objectives may omit critical physical and task structure while retaining control-irrelevant visual redundancy. We call this mismatch between visual richness and control utility the action-sufficiency gap. We investigate whether this gap can be bridged by guiding intermediate features to preserve three control-relevant structure in robotic manipulation: geometry governing motion feasibility, affordance encoding instruction-relevant entities, and goals grounding instructions in task-relevant regions. To this end, we present GIFT (Guided Intermediate Feature Training), an architecture-flexible framework for learning intermediate features that translates these structures into training-time constraints through geometry alignment, affordance prediction, and goal-region reconstruction. We instantiate GIFT in a Vision-Language-Action (VLA) policy, a direct-action World-Action Model (WAM), and an inverse-dynamics WAM while retaining each model's action formulation. Under zero-shot transfer to LIBERO-Plus, GIFT-VLA, GIFT-WAM-Fast, and GIFT-WAM-IDM outperform StarVLA-OFT, Fast-WAM, and Fast-WAM-IDM by 4.6, 12.6, and 5.2 points, reaching 79.6%, 72.6%, and 87.8%, respectively. On RoboCasa, the three GIFT variants reach 61.4%, 83.6%, and 82.3%, outperforming their counterparts by 12.6, 9.0, and 8.4 points, respectively. Together, these results establish learning functionally structured intermediate features as a reusable principle across model-specific action formulations, with especially large gains on articulated-object tasks and high-precision real-world manipulation under unseen visual and spatial perturbations. Project page: https://openphoenix-team.github.io/GIFT-pages.

Dense self-supervised learning (SSL) is a powerful paradigm for learning without annotations the local descriptors required to solve dense medical imaging tasks. We present Pix2Rep-v2, a framework for SSL of pixel- and voxel-level representations suitable for few-shot downstream applications. Pix2Rep-v2 addresses the main challenges of dense SSL by leveraging a redundancy reduction objective at the pixel-level with a principle of equivariance of dense representations, that scales efficiently to 3D or wide field-of-view applications. We evaluate our method on four datasets, across multiple tasks, multiple modalities and anatomical structures using multiple backbones in 2D and 3D, and under various data regimes. As an alternative to linear probing or full fine-tuning on the downstream task, we also propose an in-context variant, without downstream training, based on a dense prototype approach. Pix2Rep-v2 shows substantially higher data-efficiency in few-shot scenarios compared to fully supervised baselines, and is competitive with the state-of-the-art e.g., +9.3 Dice points in one-shot segmentation on the M&Ms-2 dataset. Our code and pre-trained models are publicly available at https://github.com/BioMedTP/pix2rep-v2.

Text-rich visual inputs require models that can read, retrieve, and compress language directly in pixel space, yet existing pixel-text encoders struggle with fixed resolution pretraining, visual shortcut learning, weak visual grounding, and multilingual visual text understanding. In this work, we investigate the fundamental design principles required for robust visual text representation learning. Through systematic controlled ablations, we identify four critical components: variable image resolutions and rendered font sizes provide spatial proxies for high-resolution document generalization; natural image-text pairs are indispensable for grounding and prevent text-only collapse; layout-aware rendering helps prevent pixel-level shortcuts; and a two-stage multilingual curriculum enables effective cross-lingual alignment. By integrating these principles into a scalable training recipe, we train Pixel Linguist II, a native-resolution vision encoder trained with on-the-fly rendering, unified contrastive grounding, and a multilingual curriculum over 280M training examples. Pixel Linguist II sets new state-of-the-art results on English, cross-lingual, and multilingual Visual STS and ViDoRe, while also enabling better MLLM downstream evaluation. Notably, Pixel Linguist II remains robust under 80% visual token compression, showing great promise for optical context compression. Our code and resources are available at https://github.com/Pixel-Linguist/Pixel-Linguist-II.

Pretrained vision embeddings are widely used to model how people appraise urban scenes, and they are validated almost entirely by how well they predict human ratings. Accurate prediction shows that an embedding contains the information needed to recover the ratings. It does not show that the embedding arranges scenes as the human visual system does, which is assumed when distances or dimensions in the embedding are read as perceptual. Using openly released EEG recorded while 63 adults viewed and rated street scenes of Berlin, we compared the neural representational geometry of the scenes with the geometry of pretrained vision models that differ in training objective and size, of simple image descriptors and of the ratings themselves, each relative to a noise ceiling given by the agreement between participants. No feature space reached more than about half of the noise ceiling, which corresponds to about a fifth of the reliable variance in the neural geometry. A descriptor of oriented edge energy reached the level of most pretrained models while capturing a different part of the neural geometry, and deeper layers corresponded to later neural responses. The same embeddings predicted held-out ratings well, up to r = 0.87, yet prediction accuracy and neural correspondence were not reliably related across models, and the best predictor was among the least aligned. Predicting how a street is appraised is therefore weak evidence that a model represents the street as the brain does.

We introduce a self-supervised framework for learning joint visuospatial representations from RGB-D observations. While modern vision foundation models are trained almost exclusively on RGB images, many embodied systems have access to explicit depth sensing, which provides geometric information that monocular inputs cannot recover. Our method integrates depth-derived geometric priors with a visual backbone through inter-patch and intra-patch fusion, enabling the model to encode both appearance and spatial structure efficiently. The resulting representation shows promising improvements on 3D awareness while preserving semantic transfer: it outperforms prior methods of comparable scale on multiple 3D geometry benchmarks, and remains competitive when probed for standard RGB-D semantic segmentation tasks.

Foundational visual features such as DINO have played a critical role across modern computer vision, and have recently become key components in multi-view feed-forward geometry estimators. In this work, we demonstrate that by re-distilling these multi-view models---their internal knowledge of 3D geometry---into a single-view estimator, we can obtain enhanced 3D consistent foundational features. Our key idea is to construct a multi-view teacher by fusing pretrained 2D foundation features with multi-view geometric features, and refining the fused representation with a discriminative ranking objective. Through our discriminative distillation framework, we enforce the learned features to be both 3D consistent and locally distinctive, while keeping them aligned with the feature space of the original foundation model to preserve the semantic structure of the pretrained representation. Consistency and local discriminability are critical for 3D computer vision problems such as forming semantic and geometric correspondences across images. To demonstrate the effectiveness of our method, we perform comprehensive experiments spanning multiple angles: direct feature analysis, dense prediction transfer, and explicit 3D lifting and rendering. Across these evaluations, our method consistently produces stronger 3D-aware foundation features that improve multi-view consistency and local discriminability while preserving the semantic transferability of the original representation.

Semantic vision encoders have become a central visual interface for multimodal understanding and semantic conditioning in image generation. However, their final tokens discard fine-grained visual details, leading to poor pixel reconstruction and limiting their use in reconstruction-sensitive tasks such as image generation and editing. In this work, we ask whether understanding, generation, and editing can be modeled in a single visual representation space built from a pretrained semantic ViT. We show that the frozen Transformer blocks of a semantic ViT are not intrinsically unable to preserve visual details. Instead, the original patch parameterization drives the representation toward semantic abstraction, making fine-grained information difficult to recover from the final tokens. Based on this observation, we introduce \emph{Patch Reparameterization}, which preserves the original semantic pathway while adding a reconstruction-aware patch embedding that provides fine-grained visual information to the same frozen ViT blocks. The resulting unified representation preserves multimodal understanding while enabling high-fidelity image reconstruction and a favorable reconstruction--generation trade-off. We further scale this representation into \emph{UniSpace}, an 8B Mixture-of-Transformer-Experts model that performs understanding, generation, and editing in the same visual space without a separate VAE pathway. System-level evaluations demonstrate practical text-to-image generation and instruction-based image editing, showing that a reparameterized pretrained ViT can serve as a unified visual interface for scalable multimodal modeling.

Compositional analysis of frozen vision encoders should determine both what changed and where it changed. Standard factor probes score these axes separately, however, and can reward multiple operations that reuse the same predicted slot. We call this failure operation laundering. We introduce an injectively aligned leave-one-cell-out protocol over support x operation grids and SO-OPF, a readout that factors cell energy into support salience and a competitive operation posterior. This formulation separates two questions that aggregate scores conflate: whether the carrier composes held-out bindings when the grid is known, and whether that grid can be recovered from flat cell labels. With frozen DINOv3 features, known factorial assignment reaches 0.874 injective accuracy on Shapes3D-Extended and 0.799 on globally image-disjoint COCO; learning the assignment from flat labels reaches 0.769 and 0.762, respectively. Under matched-axis-aware supervision on Shapes3D, the factored carrier improves learned-assignment accuracy from 0.653 to 0.841 over a dense carrier and eliminates its laundering gap. SigLIP2 replicates the COCO separation. A rebuilt MuJoCo substrate exposes a boundary: learned-assignment accuracy is 0.569 with DINOv3 and 0.484 with SigLIP2, with substantial slot collapse. Thus factored readout and injective evaluation recover held-out bindings on two substrates while exposing, rather than hiding, a renderer-specific failure boundary; they do not establish universal recovery from flat labels.

Composition, the deliberate arrangement of visual elements, is central to how meaning, emotion, and aesthetic quality are conveyed in artwork, yet it remains among the least formalized dimensions of visual understanding. Prior work highlights a persistent gap in learning meaningful compositional representations, attributing it to semantic bias and suggesting that human-inspired approaches may be key. We compare two parallel paradigms for composition analysis: a human-inspired method grounded in perceptual grouping, and fine-tuned foundation models enabled by recent large-scale compositional datasets. The human-inspired approach uses object-centric models for region-level decomposition and a graph attention network to capture spatial relationships between elements. Both paradigms are evaluated on composition score/category prediction, compositional image retrieval, and visual saliency detection. With frozen encoders, the human-inspired method achieves competitive performance while remaining interpretable. When sufficient data enables fine-tuning, large self-supervised models outperform significantly, but at the cost of interpretability and cross-domain generalization.

Vision transformers (ViTs) have become the de facto standard for image encoding across many perception tasks. Despite their empirical success, it remains mechanistically unclear how they encode low-level features, given their lack of inductive biases: ViTs process information globally rather than relying on local structure. Biological visual systems, in contrast, build low-level features, such as orientation selectivity in the primary visual cortex, by combining information from small, localized regions of the visual field. These features are general-purpose representations, shared and required across multiple specialized neural pathways, unlike higher-level, task-specific semantic features. This raises the question if such biologically-grounded features arise in ViTs. In this work, we systematically study how orientation selectivity emerges in ViTs by introducing a suite of neuroscience-inspired metrics: representational similarity score (RSS), orientation recruitment score (ORS), and orientation tuning bandwidth to quantify how orientation is encoded in representational geometry and as a function of model depth. Through extensive analysis, we find that: (1) the training paradigm is the strongest determinant of orientation selectivity, with models sharing an objective, peaking at comparable relative depths regardless of scale (2) many units are orientation-selective early in training, with early-to-middle layers recruiting more such units over time, while deeper layers lose selectivity and broaden their tuning toward semantic encoding and (3) our metrics offer a mechanistic heuristic for how many layers to unfreeze for best downstream generalization. Our framework presents a way to track biologically-grounded features during ViT training, probes how desired properties are encoded in transformer representations, and builds a systematic understanding of how ViTs generalize across tasks.

Visual document retrieval has recently become increasingly important in applications such as enterprise search, scientific literature discovery, and retrieval-augmented generation. These applications depend on efficiently identifying query-relevant pages across large collections of visually rich documents. Existing methods commonly adopt late-interaction architectures that encode and index documents offline to enable scalable and low-latency online retrieval. Despite its efficiency, this paradigm requires each document to be encoded into a fixed representation before the query is known. However, the same content in a visual document may induce different interpretations depending on the query intent, which a fixed representation struggles to capture. Yet postponing document encoding until the query arrives would incur prohibitive online retrieval latency. To address this gap, we propose VaRS-Doc, a visual document retrieval framework that diversifies document representations by enabling the model to actively explore variant latent interpretations during document encoding, while preserving efficient late-interaction retrieval in which each query adaptively selects the best-fit representation. We further introduce a two-stage training strategy that encourages the model to capture complementary semantic interpretations and prevents it from falling back to train a single dominant representation. Experiments on visual document retrieval benchmarks show that VaRS-Doc achieves state-of-the-art retrieval performance, offering a practical solution to the mismatch between query-agnostic document encoding and query-specific retrieval needs. Code is available at https://github.com/bokufa/VaRS-Doc.

Frozen vision foundation models are commonly evaluated through a single global image embedding, but this interface can conflate missing information with information lost at readout time. We study this distinction by keeping a pretrained vision encoder frozen and varying only the readout applied to its final patch tokens. We compare standard global readouts against a lightweight foveated readout, which attention-pools patch tokens using a learned or question-conditioned query, and against an oracle readout with access to the annotated target region. We evaluate these interfaces on three localized binding problems: a controlled synthetic color--shape binding task under clutter, a color-free crowded shape-detection variant, and a GQA-derived natural-image task where paired questions ask for the colors of different same-category objects in the same image. Global readouts perform near perfectly when the synthetic target appears alone, but collapse under clutter and counterfactual target edits, whereas the foveated readout recovers most of the oracle-accessible signal. On the GQA-derived task, question-independent global image vectors improve only modestly over question-only priors, while question-conditioned foveation substantially improves paired localized color accuracy. A counterfactual nuisance-to-signal ratio explains the synthetic failures: global pooling dilutes localized label-changing evidence while exposing the probe to nuisance variation from irrelevant objects. These results indicate that apparent spatial blindness in frozen vision models can arise from the global embedding interface rather than from an absence of spatial information in the frozen patch tokens.

Optical microscopy of particle and fiber dispersions involves interpreting subtle visual cues influenced by specimen morphology, chemical composition, magnification, and illumination conditions. We introduce an artificial intelligence (AI) distillation framework that extracts semantically rich image embeddings from microscopy images using semantic anchors. A multimodal teacher combines each image's visual embedding with three text embeddings representing illumination modality, magnification, and specimen identity and morphology. Generated by LongCLIP's extended-context text encoder, this yields a 2304-dimensional block-structured teacher vector whose component blocks remain physically interpretable throughout training and inference. A student vision transformer (ViT) with a multi-layer perceptron (MLP) decoder is trained to reconstruct this teacher vector from the image alone, minimizing a mean absolute error (L1) loss that enforces coordinate-level fidelity to the teacher's block structure. A cross-entropy term over pseudo-classes derived from HDBSCAN clustering of the teacher embedding space acts as a collapse-prevention regularizer, enforcing inter-cluster separation without requiring contrastive negative mining. At inference, the student operates on image input alone, producing compact embeddings that recover the full semantic content of the teacher vector. The framework achieves approximately 80% pseudo-class validation accuracy and 75% Recall@1 on fine-grained specimen description labels under leave-one-out nearest-neighbor retrieval. These results demonstrate that semantic anchoring enables a vision-only student to acquire richer and more interpretable representations than image-only training, with direct applicability to retrieval, classification, and exploratory analysis of heterogeneous particle and fiber dispersions.

Visual imitation learning enables robots to acquire visuomotor skills directly from images, yet RGB observations lack explicit geometric cues, making learned policies brittle to camera perturbations. To address this, we propose \textbf{Ray-conditioned Vision Transformer Encoder (RayViT)}, a lightweight architecture that injects camera geometry into pretrained ViT backbones. RayViT represents camera geometry as a Plücker ray map, patchifies it into ray features, and uses gated cross-attention to produce a ray-conditioned class token. These ray features are added as dense positional embeddings, while the ray class token replaces the original ViT class token to provide a geometry-aware summary representation. We combine this approach with an auxiliary cosine similarity loss to consistently improve the performance and robustness for geometry-aware tokens. Experiments on sim- and real-robot tasks demonstrate that RayViT improves robustness by approximately 13 percentage points under camera perturbations in multi-task RoboCasa benchmark and by 1.78 average completed stages in real-world multi-task success rate compared to baselines.

Design and architectural archives encode expert human knowledge in graphical formats, providing a critical testbed for design-inspired Machine Learning (ML) challenges absent with typical computer vision benchmarks. Building on JONES-19, a small-size image dataset based on The Grammar of Ornament (London, 1857), we evaluate the discriminative performance of Convolutional Neural Networks (CNNs) in two model training strategies: (a) ImageNet pretraining for domain-general "visual common sense," and (b) learning from scratch on the design data in JONES-19. We find that while domain-general priors improve discriminative performance, learning from scratch augmented with repeated local sampling (multi-crop) effectively recovers these gains. For highly structured design data, local design-driven representations provide sufficient foundation for learning, challenging a reliance on massive general-purpose pretraining. These findings suggest that in specialized design domains, careful curation of smaller high-quality datasets that capture empirical and formal design principles may prove more effective and informative on the nature of a particular design domain than prioritizing large-scale data collection.

Current deep learning-based character vision studies, e.g., text recognition, character image denoising, and historical text completion, are offering new solutions for learning, managing, and utilizing character resources. However, the performance of these studies peaks only with large and balanced datasets, which is a rarity with real-world character datasets, especially for logographic character languages, e.g., Chinese. The imbalance in data distribution of logographic characters is a common issue due to differences in character usage frequency and new characters being continuously created. In this paper, we propose a novel method for logographic character recognition, which introduces a multi-modal learning approach using visual semantics and contextual semantics of characters. A novel pre-training strategy is designed to enhance deep visual representations, especially for datasets suffering from issues of imbalanced and rare instances, by extracting the contextual semantics of each character from the corresponding language models. We conduct experiments across various datasets to evaluate our character recognition method and further validate the contrastive pre-training strategy by several downstream tasks. Experimental results demonstrate the superiority of our method compared to state-of-the-art methods.

Visible-infrared (VIS-IR) alignment is a key pre-training task for robust multi-sensor perception. Most existing methods use uniform patch-wise contrastive learning, but this can be unreliable in VIS-IR data because imaging-physics differences make some spatially paired regions inherently less comparable, and aligning them with equal strength hinders representation learning and downstream transfer. In this paper, we revisit VIS-IR pre-training from a sampling perspective and propose Importance-Aware Sampling (IAS), which adjusts training emphasis based on patch reliability. Specifically, IAS (i) derives patch weights from infrared structural cues and uses them to reweight the contrastive objective; (ii) learns a soft importance mask with a lightweight sampler, optionally warm-started from the hand-crafted prior; and (iii) employs a patch curriculum learning strategy that gradually expands from high-reliability regions to harder patches. It is worth noting that IAS is plug-and-play and works with both patch-/correlation-level alignment (e.g., UNIV-style) and image-level contrastive baselines (e.g., ImageBind-style). Extensive experiments on multiple VIS-IR benchmarks demonstrate consistent improvements over strong baselines, including for IR semantic segmentation, IR object detection and VIS semantic segmentation and cross-modal retrieval task. Code will be released on https://github.com/KlayMa527/IAS.

Functional magnetic resonance imaging (fMRI)-based visual decoding aims to recover visual information from measured brain activity, commonly by mapping fMRI responses into latent visual features for downstream decoding tasks. Most existing methods learn mappings from fMRI responses to visual features extracted by artificial neural networks (ANNs), yet it remains unclear whether ANN-derived features provide suitable targets for brain decoding. In this study, we investigate spiking neural network (SNN)-derived visual features as alternative targets for fMRI-based visual decoding. We compare an ANN baseline with four SNN variants from the same architectural family, which differ in their spiking dynamics. To isolate the effect of the target features, all models use the same L2-regularized linear fMRI-to-feature decoder, while only the feature vectors used as regression targets are varied. Compared with the ANN baseline, SNN-derived features exhibit stronger alignment with fMRI responses and improve visual semantic decoding performance. For instance, on the GoD dataset, SNN-derived features reduce feature-prediction error from 0.7707 to 0.0282 and improve top-1 semantic decoding accuracy from 0.1800 to 0.4400. Ablation results further indicate that both spiking neural dynamics and temporal simulation steps contribute to the observed advantage. These findings support SNN-derived features as effective brain-decodable visual representations and highlight target feature design as an important component of fMRI-based visual decoding.

Single-cell light microscopy images have become an important data source for characterizing cell phenotypes, but their complexity and heterogeneity pose challenges to high-throughput automated analysis. Existing representation learning methods mostly rely on task-oriented modeling, which is limited by specific datasets and predefined tasks, making them difficult to generalize across different cell types and microscopy modalities, and experimental conditions. Although general-purpose methods have improved the generalization ability of image representation in recent years, their limited utilization of experimental background and biological context information still poses challenges in complex phenotypic analysis. Here, we propose scMIR, a vision-language foundation model for single-cell light microscopy image representation. By synergistically combining self-supervised image reconstruction with text-guided cross-modal alignment, scMIR can simultaneously encode morphological and biological semantic information in a unified representation space. scMIR is pre-trained on 207,957 image-text pairs, covering various cell types, microscopy modalities, and perturbation conditions. scMIR outperforms existing general models and task-oriented methods as systematically evaluated on various complex tasks using 16 benchmark datasets, including cell classification, clustering, phenotype inference, and batch effect correction tasks. Furthermore, scMIR shows a strong generalization ability across various tasks without requiring task-specific fine-tuning. With its unique advantages, we envision scMIR may promote the standardization and automation of high-throughput phenotyping workflows through supporting various downstream analysis tasks.

Vision Mamba models replace quadratic self-attention with linear complexity selective state space models (SSMs), emerging as efficient visual backbones. However, MambaOut demonstrates that a Gated CNN block can match or exceed VMamba on image classification, questioning the necessity of SSMs for vision. This raises a fundamental question: do VMamba and MambaOut encode visual information differently at the representation level? To investigate, we apply cross model centered kernel alignment (CKA) analysis and find that VMamba's final stage blocks form representations distinctly different from both MambaOut and its own preceding blocks. We therefore focus on the final block features, decomposing each spatial token into magnitude and direction. MambaOut concentrates class-discriminative information in high-norm foreground tokens that align with Grad-CAM attribution. VMamba, by contrast, produces high-norm tokens predominantly in background regions, misaligned with Grad-CAM, yet preserves discriminative signals primarily in token directions. These observations reveal that the two models rely on different encoding strategies. We connect this difference to high-resolution classification and semantic segmentation. VMamba distributes logit support broadly across object regions, whereas MambaOut relies on sparse dominant tokens, a strategy that becomes less stable as token counts grow. Under full fine-tuning for segmentation, VMamba consistently outperforms MambaOut. These results suggest that VMamba's advantage in dense prediction stems not merely from the SSM mechanism or sequence length, but from how semantic evidence is organized across token magnitude, direction. Ultimately, we conclude that token magnitude and directional structure serve as critical axes for improving visual backbones, particularly under dense supervision.

Deep neural networks trained on natural images are shown to produce outputs consistent with human observers for brightness illusions. While this phenomenon has been documented across architectures, all evidence, to date, is measured at the output level: restored pixels, decoded trajectories, or classification decisions. Whether these models actually represent illusions internally, and if so where and how, remains unknown. We show that denoising models develop illusion-sensitive representations at specific internal layers, across varied architectures. Specifically, we identify the layers and channels that discriminate illusory from physically matched control regions. We show that the denoising objective is a more important driver of the effect than the architecture. On domain-appropriate stimuli, these activations track a validated psychophysical model of human brightness perception (FLODOG; Spearman $ρ\geq 0.70$) and scale monotonically with parametric illusion strength. Leveraging these findings, we provide causal evidence via channel ablation showing that illusion-sensitive channels specifically and substantially affect the internal signal. Yet injecting these representations into the generation pipeline produces no measurable pixel shift across all tested architectures; we term such representations perceptual phantoms: active in internal processing yet invisible to any output-based evaluation. While related internal-output dissociations have been characterized in language models, this is the first such characterization for perceptual representations in denoising vision models.

Do vision models see colors the way humans do? Existing evaluations of color representations usually compare them with geometric spaces such as CIELAB or with discrete color labels. These references capture perceptual distance or category membership, but not the graded way in which people organize colors. We evaluate color grounding against a fuzzy perceptual model with 86 graded categories fitted to human survey data. The framework can be applied to any image encoder and measures three complementary properties: category boundaries, category compactness, and graded alignment beyond what color geometry alone can explain. Across eleven Vision Transformer encoders, the category-level results are broadly similar, whereas graded alignment differs substantially. Masked Autoencoders achieve the strongest beyond-geometry alignment, with confidence intervals that do not overlap those of the other encoders. A layer-wise analysis further shows that masked reconstruction preserves this structure toward the output. On natural images, MAE represents surface color globally, while language-supervised models encode color more strongly in relation to the foreground object. These results show that human-like color grounding has several distinct aspects that should not be reduced to a single score.

Automatic writer identification systems have progressed remarkably in recent years, yet their deployment in archival paleography remains limited by the scarcity of labeled training data, open scribe sets, and degraded image quality. We present TextileNet, a fully convolutional multi-task network trained exclusively on synthetic data to produce dense pixel-level texture embeddings, which we transfer zeroshot to historical manuscript analysis. As an original contribution to evaluation methodology, we designed a paleographic visual quiz of 80 pair and triplet questions and administered it to a range from lay participants to senior paleographers under strict anonymity, establishing to our knowledge for the first time a human baseline for script-style discrimination on late medieval text. We employ TextileNet embeddings to perform zero-shot retrieval on sub-word granularity for hand and gender identification. Our experimental results help in building the credibility of TextileNet in the paleographic domain, but more than that demonstrate in experimental terms that the question of gender in handwriting needs to be treated with caution.

Fusing standard RGB frames with asynchronous event streams has emerged as a definitive paradigm for robust perception in degraded environments. Although unified backbones have recently gained traction in multi-modal vision, adapting them to the RGB-Event domain remains fundamentally challenging. Existing architectures either resort to decoupled dual encoders that double computational overhead, or adopt generic unified designs that fail to resolve implicit geometric parallax and cross-spectral aliasing under the extreme representational divide between dense intensity grids and sparse kinematic spikes. To transcend these bottlenecks, we present Evita, the first unified backbone specifically engineered for dedicated dense RGB-Event parsing. To achieve profound modal synergy, Evita explicitly embeds a suite of intrinsic co-learning modules directly into every encoder layer. Specifically, it features Geometric Parallax Rectification for adaptive spatial alignment, Harmonic Spectral Resonance for texture transfer exclusively in the complex frequency domain, and Transient Global Routing for event-driven asymmetric attention. To guarantee robust feature extraction against spatial misalignments and decouple representations from specific event encodings, we construct N-ImageNetV2 alongside a stochastic event representation mixing pretraining protocol, empowering the network to seamlessly accommodate arbitrary event formats in downstream tasks. Extensive evaluations across the DELIVER, DDD17, and DSEC benchmarks confirm that Evita establishes new state-of-the-art metrics while delivering a superior accuracy-latency trade-off for real-time multimodal perception.The code are publicly available at: https://github.com/chaineypung/Evita.

Figure-ground organization in the human visual system relies on several shape-based cues, including surroundedness, convexity, and symmetry. While these cues have been extensively studied using abstract stimuli, little is known about how they operate under natural conditions or how they arise from the statistics of natural scenes. Deep neural networks offer a promising path forward: a model that relies on the same figure-ground cues as humans would provide tractable experimental access to the underlying mechanisms. In this study, we evaluate shape-based figure-ground organization in Vision Transformers (ViTs), for which prior work has demonstrated the emergence of object-based grouping. We test 25 ViTs spanning supervised and self-supervised training objectives, by fitting linear probes to predict figure-ground assignment from intermediate patch representations using both natural images and controlled artificial stimuli that isolate individual cues. Our results show that ViTs robustly encode surroundedness and convexity, and that probes trained on natural images generalize zero-shot to artificial stimuli across several models. For symmetry we observe mixed results: the cue is encoded for uniformly colored but not for textured regions. Taken together, our findings demonstrate that Gestalt-like figure-ground cues can be learned from natural scene statistics and position ViTs as a compelling model system for studying the computational mechanisms of perceptual organization. Code and data is available at https://github.com/mtangemann/mlvbench

In computational vision science, Convolutional Neural Networks (CNNs) have emerged as a popular model of biological vision because of the alignment they can exhibit with neural and behavioral data in humans and animals. However, it remains unclear to what extent this alignment persists for visual tasks that extend beyond the canonical object recognition paradigm based on well defined semantic content. In this study, we diverge from the common object-centric view by focusing on another aspect of vision: texture perception. We consider textures of different complexity generated with three different algorithms from the same source images. Using a rank-based statistic, we quantify the information encoded in the internal representations of a CNN and three Vision Transformers (ViTs), and we compare the similarity of these representations to those inferred from human psychophysics data. We find that the representation of textures is aligned in different ViTs, but not between the ViTs and the CNN; that ViTs form similar representations for textures of different complexity; that human performance in recognizing textures can be better predicted from ViTs representations rather than CNN representations. Taken together, these results suggest that ViTs may capture more faithfully than CNNs how texture patterns are visually processed by humans, and that the representations of texture stimuli in computational models may be driven by the network architecture.

This paper studies the problem of learning a joint distribution from marginal observations, which is inherently ill-posed due to the ambiguity of feasible couplings. We propose LUD-MSR, a latent-variable probabilistic framework that models the joint distribution via auxiliary representations and optimizes evidence lower bounds using only marginal data. Under mild assumptions, we establish an upper bound on the distribution approximation error. This analysis reveals a trade-off in representation learning between domain consistency and information preservation. To address this trade-off, we introduce a Multi-Scale image Representation (MSR) mapping that exploits structural similarity at coarse scales while suppressing domain-specific variations. We show that MSR achieves a more favorable balance of this trade-off compared to existing approaches. Experiments on real-world denoising benchmarks, including cryo-electron microscopy (cryo-EM), demonstrate the effectiveness of the proposed framework.

Rendering time series as chart images for CNN-based classification has become increasingly common in time-series classification (TSC). However, it remains unclear whether models learn underlying temporal patterns or rely on encoding-specific visual cues introduced by chart design. We present VEIL: a systematic study examining how chart encodings influence learned representations through complementary analyses of similarity, transferability, and attribution. Attention-guided training appears to mitigate this effect when encoding sensitivity is consistently identified across diagnostics, but provides limited or negative benefit when such signals are absent. These findings position VEIL within the broader question of how machines perceive visualizations -- extending graphical perception from human readers to vision models -- and show that visualization design choices shape learned representations in ways that warrant treating chart-based TSC as a representation and measurement problem rather than a simple modeling decision.

Low-light degradation reduces image visibility and weakens structural cues that are important for visual representation and scene understanding. Existing low-light image enhancement methods mainly focus on appearance restoration, while insufficiently exploiting scene geometry to preserve structural consistency. To address this limitation, this paper proposes a Depth-guided Multi-scale Attention Network (DMSA-Net) for geometry-aware low-light image enhancement. DMSA-Net introduces depth-related structural priors into low-light representation learning through reflectance-geometry interaction. A Retinex-based decomposition module is first used to obtain illumination-invariant reflectance representations, from which depth cues are inferred to characterize scene structure under degraded illumination. A multi-scale depth-guided fusion strategy is then embedded into a hierarchical encoder-decoder architecture, where depth-aware attention adaptively integrates geometric and appearance features. Experiments on several benchmark datasets show that DMSA-Net achieves effective low-light restoration while improving structural preservation. Moreover, we construct LOL-D, a depth-augmented low-light dataset, to facilitate research on geometry-aware low-light vision.