Object-Centric Representation Learning

The de facto approach in video object-centric learning maintains temporal consistency through learned dynamics modules that predict future object representations, called slots. We demonstrate that these predictors function as expensive approximations of discrete correspondence problems. Modern self-supervised vision backbones already encode instance-discriminative features that distinguish objects reliably. Exploiting these features eliminates the need for learned temporal prediction. We introduce Grounded Correspondence, a framework that replaces learned transition functions with deterministic bipartite matching. Slots initialize from salient regions in frozen backbone features. Frame-to-frame identity is maintained through Hungarian matching on slot representations. The approach requires zero learnable parameters for temporal modeling yet achieves competitive performance on MOVi-D, MOVi-E, and YouTube-VIS. Project page: https://magenta-sherbet-85b101.netlify.app/

Recent EEG-to-image retrieval methods leverage pretrained vision encoders and foveation-inspired priors, but typically assume a fixed, center-focused view. This center bias conflicts with content-driven human attention, creating a geometric-semantic dissociation between visual features and EEG responses. We propose SIMON, a saliency-aware multi-view framework for zero-shot EEG-to-image retrieval. SIMON combines foreground segmentation and saliency prediction to select fixation centers via Saliency-Aware Sampling (SAS), then generates foveated views that emphasize informative object regions while suppressing background clutter. On THINGS-EEG, SIMON achieves state-of-the-art performance in both intra-subject and inter-subject settings, reaching an average Top-1 accuracy of 69.7% and 19.6%, respectively, consistently outperforming recent competitive baselines. Analyses across sampling granularity, EEG channel topology, and visual/brain encoder backbones further support the robustness of saliency-aware multi-view integration. Our code and models are publicly available at https://github.com/simonlink666/SIMON.

Human visual perception offers valuable insights for understanding computational principles of motion-based scene interpretation. Humans robustly detect and segment moving entities that constitute independently moveable chunks of matter, whether observing sparse moving dots, textured surfaces, or naturalistic scenes. In contrast, existing computer vision systems lack a unified approach that works across these diverse settings. Inspired by principles of human perception, we propose a generative model that hierarchically groups low-level motion cues and high-level appearance features into particles (small Gaussians representing local matter), and groups particles into clusters capturing coherently and independently moveable physical entities. We develop a hardware-accelerated inference algorithm based on parallelized block Gibbs sampling to recover stable particle motion and groupings. Our model operates on different kinds of inputs (random dots, stylized textures, or naturalistic RGB video), enabling it to work across settings where biological vision succeeds but existing computer vision approaches do not. We validate this unified framework across three domains: on 2D random dot kinematograms, our approach captures human object perception including graded uncertainty across ambiguous conditions; on a Gestalt-inspired dataset of camouflaged rotating objects, our approach recovers correct 3D structure from motion and thereby accurate 2D object segmentation; and on naturalistic RGB videos, our model tracks the moving 3D matter that makes up deforming objects, enabling robust object-level scene understanding. This work thus establishes a general framework for motion-based perception grounded in principles of human vision.

While current deep learning models achieve high performance by learning statistical correlations from vast datasets,which stands in stark contrast to human learning. They lack the flexibility of humans-particularly preverbal infants-to autonomously acquire the underlying structure of the world from limited experience and adapt to novel situations. In this study, we propose an unsupervised representation learning method based on a hierarchical relationship in group operations, rather than statistical independence, aiming to build a computational model of the cognitive development of infants. The proposed model features an integrated architecture that simultaneously performs object segmentation and the extraction of motion laws from dynamic image sequences. By introducing the Homomorphism from algebra as a structural constraint within a neural network, the model structurally separates pixel-level changes into meaningful, decomposed transformation components, such as translation and deformation. Using interaction scenes (chasing and evading tasks) based on developmental science findings, we experimentally demonstrate that the model can segment multiple objects into individual slots without any ground-truth labels. Furthermore, we confirmed that relative movements between objects, such as approaching or receding, are accurately mapped and structured into a one-dimensional additive latent space. These results suggest that by introducing algebraic geometric constraints rather than relying solely on statistical correlation learning, physically interpretable "disentangled representations" can be acquired. This study contributes to the understanding of the process by which infants internalize environmental laws as structures and provides a new perspective for constructing artificial systems with developmental intelligence.

While foundation models drive steady progress in image segmentation and diffusion algorithms compose always more realistic images, the seemingly simple problem of identifying recurrent patterns in a collection of images remains very much open. In this paper, we focus on sprite-based image decomposition models, which have shown some promise for clustering and image decomposition and are appealing because of their high interpretability. These models come in different flavors, need to be tailored to specific datasets, and struggle to scale to images with many objects. We dive into the details of their design, identify their core components, and perform an extensive analysis on clustering benchmarks. We leverage this analysis to propose a deep sprite-based image decomposition method that performs on par with state-of-the-art unsupervised class-aware image segmentation methods on the standard CLEVR benchmark, scales linearly with the number of objects, identifies explicitly object categories, and fully models images in an easily interpretable way.

Robust robotic manipulation requires not only predicting how the scene evolves over time, but also recognizing task-relevant objects in complex scenes. However, existing VLA models face two limitations. They typically act only on the current frame, while future prediction and object-aware reasoning are often learned in separate latent spaces. We propose OFlow (injecting Object-Aware Temporal Flow Matching into VLAs), a framework that addresses both limitations by unifying temporal foresight and object-aware reasoning in a shared semantic latent space. Our method forecasts future latents with temporal flow matching, factorizes them into object-aware representations that emphasize physically relevant cues while filtering task-irrelevant variation, and conditions continuous action generation on these predictions. By integrating OFlow into VLA pipelines, our method enables more reliable control under distribution shifts. Extensive experiments across LIBERO, LIBERO-Plus, MetaWorld, and SimplerEnv benchmarks and real-world tasks demonstrate that object-aware foresight consistently enhances robustness and success.

Large Multimodal Models (LMMs) have achieved remarkable progress in general-purpose vision--language understanding, yet they remain limited in tasks requiring precise object-level grounding, fine-grained spatial reasoning, and controllable visual manipulation. In particular, existing systems often struggle to identify the correct instance, preserve object identity across interactions, and localize or modify designated regions with high precision. Object-centric vision provides a principled framework for addressing these challenges by promoting explicit representations and operations over visual entities, thereby extending multimodal systems from global scene understanding to object-level understanding, segmentation, editing, and generation. This paper presents a comprehensive review of recent advances at the convergence of LMMs and object-centric vision. We organize the literature into four major themes: object-centric visual understanding, object-centric referring segmentation, object-centric visual editing, and object-centric visual generation. We further summarize the key modeling paradigms, learning strategies, and evaluation protocols that support these capabilities. Finally, we discuss open challenges and future directions, including robust instance permanence, fine-grained spatial control, consistent multi-step interaction, unified cross-task modeling, and reliable benchmarking under distribution shift. We hope this paper provides a structured perspective on the development of scalable, precise, and trustworthy object-centric multimodal systems.

Recent works on 3D scene understanding leverage 2D masks from visual foundation models (VFMs) to supervise radiance fields, enabling instance-level 3D segmentation. However, the supervision signals from foundation models are not fundamentally object-centric and often require additional mask pre/post-processing or specialized training and loss design to resolve mask identity conflicts across views. The learned identity of the 3D scene is scene-dependent, limiting generalizability across scenes. Therefore, we propose a dataset-level, object-centric supervision scheme to learn object representations in 3D Gaussian Splatting (3DGS). Building on a pre-trained slot attention-based Global Object Centric Learning (GOCL) module, we learn a scene-agnostic object codebook that provides consistent, identity-anchored representations across views and scenes. By coupling the codebook with the module's unsupervised object masks, we can directly supervise the identity features of 3D Gaussians without additional mask pre-/post-processing or explicit multi-view alignment. The learned scene-agnostic codebook enables object supervision and identification without per-scene fine-tuning or retraining. Our method thus introduces unsupervised object-centric learning (OCL) into 3DGS, yielding more structured representations and better generalization for downstream tasks such as robotic interaction, scene understanding, and cross-scene generalization.

World models that predict future states from video remain limited by flat latent representations that entangle objects, ignore causal structure, and collapse temporal dynamics into a single scale. We present HCLSM, a world model architecture that operates on three interconnected principles: object-centric decomposition via slot attention with spatial broadcast decoding, hierarchical temporal dynamics through a three-level engine combining selective state space models for continuous physics, sparse transformers for discrete events, and compressed transformers for abstract goals, and causal structure learning through graph neural network interaction patterns. HCLSM introduces a two-stage training protocol where spatial reconstruction forces slot specialization before dynamics prediction begins. We train a 68M-parameter model on the PushT robotic manipulation benchmark from the Open X-Embodiment dataset, achieving 0.008 MSE next-state prediction loss with emerging spatial decomposition (SBD loss: 0.0075) and learned event boundaries. A custom Triton kernel for the SSM scan delivers 38x speedup over sequential PyTorch. The full system spans 8,478 lines of Python across 51 modules with 171 unit tests. Code: https://github.com/rightnow-ai/hclsm

Self-supervised Vision Transformers (ViTs) like DINO show an emergent ability to discover objects, typically observed in [CLS] token attention maps of the final layer. However, these maps often contain spurious activations resulting in poor localization of objects. This is because the [CLS] token, trained on an image-level objective, summarizes the entire image instead of focusing on objects. This aggregation dilutes the object-centric information existing in the local, patch-level interactions. We analyze this by computing inter-patch similarity using patch-level attention components (query, key, and value) across all layers. We find that: (1) Object-centric properties are encoded in the similarity maps derived from all three components ($q, k, v$), unlike prior work that uses only key features or the [CLS] token. (2) This object-centric information is distributed across the network, not just confined to the final layer. Based on these insights, we introduce Object-DINO, a training-free method that extracts this distributed object-centric information. Object-DINO clusters attention heads across all layers based on the similarities of their patches and automatically identifies the object-centric cluster corresponding to all objects. We demonstrate Object-DINO's effectiveness on two applications: enhancing unsupervised object discovery (+3.6 to +12.4 CorLoc gains) and mitigating object hallucination in Multimodal Large Language Models by providing visual grounding. Our results demonstrate that using this distributed object-centric information improves downstream tasks without additional training.

Multimodal Large Language Models (MLLMs) have shown strong performance on Video Temporal Grounding (VTG). However, their coarse recognition capabilities are insufficient for fine-grained temporal understanding, making task-specific fine-tuning indispensable. This fine-tuning causes models to memorize dataset-specific shortcuts rather than faithfully grounding in the actual visual content, leading to poor Out-of-Domain (OOD) generalization. Object-centric learning offers a promising remedy by decomposing scenes into entity-level representations, but existing approaches require re-running the entire multi-stage training pipeline from scratch. We propose SlotVTG, a framework that steers MLLMs toward object-centric, input-grounded visual reasoning at minimal cost. SlotVTG introduces a lightweight slot adapter that decomposes visual tokens into abstract slots via slot attention and reconstructs the original sequence, where objectness priors from a self-supervised vision model encourage semantically coherent slot formation. Cross-domain evaluation on standard VTG benchmarks demonstrates that our approach significantly improves OOD robustness while maintaining competitive In-Domain (ID) performance with minimal overhead.

Standard vision models treat objects as independent points in Euclidean space, unable to capture hierarchical structure like parts within wholes. We introduce Worldline Slot Attention, which models objects as persistent trajectories through spacetime worldlines, where each object has multiple slots at different hierarchy levels sharing the same spatial position but differing in temporal coordinates. This architecture consistently fails without geometric structure: Euclidean worldlines achieve 0.078 level accuracy, below random chance (0.33), while Lorentzian worldlines achieve 0.479-0.661 across three datasets: a 6x improvement replicated over 20+ independent runs. Lorentzian geometry also outperforms hyperbolic embeddings showing visual hierarchies require causal structure (temporal dependency) rather than tree structure (radial branching). Our results demonstrate that hierarchical object discovery requires geometric structure encoding asymmetric causality, an inductive bias absent from Euclidean space but natural to Lorentzian light cones, achieved with only 11K parameters. The code is available at: https://github.com/iclrsubmissiongram/loco.

Video Object-Centric Learning seeks to decompose raw videos into a small set of object slots, but existing slot-attention models often suffer from severe over-fragmentation. This is because the model is implicitly encouraged to occupy all slots to minimize the reconstruction objective, thereby representing a single object with multiple redundant slots. We tackle this limitation with a reconstruction-guided slot curriculum (SlotCurri). Training starts with only a few coarse slots and progressively allocates new slots where reconstruction error remains high, thus expanding capacity only where it is needed and preventing fragmentation from the outset. Yet, during slot expansion, meaningful sub-parts can emerge only if coarse-level semantics are already well separated; however, with a small initial slot budget and an MSE objective, semantic boundaries remain blurry. Therefore, we augment MSE with a structure-aware loss that preserves local contrast and edge information to encourage each slot to sharpen its semantic boundaries. Lastly, we propose a cyclic inference that rolls slots forward and then backward through the frame sequence, producing temporally consistent object representations even in the earliest frames. All combined, SlotCurri addresses object over-fragmentation by allocating representational capacity where reconstruction fails, further enhanced by structural cues and cyclic inference. Notable FG-ARI gains of +6.8 on YouTube-VIS and +8.3 on MOVi-C validate the effectiveness of SlotCurri. Our code is available at github.com/wjun0830/SlotCurri.

In visual scene understanding tasks, it is essential to capture both invariant and equivariant structure. While neural networks are frequently trained to achieve invariance to transformations such as translation, this often comes at the cost of losing access to equivariant information - e.g., the precise location of an object. Moreover, invariance is not naturally guaranteed through supervised learning alone, and many architectures generalize poorly to input transformations not encountered during training. Here, we take an approach based on analysis-by-synthesis and factoring using resonator networks. A generative model describes the construction of simple scenes containing MNIST digits and their transformations, like color and position. The resonator network inverts the generative model, and provides both invariant and equivariant information about particular objects. Sparse features learned from training data act as a basis set to provide flexibility in representing variable shapes of objects, allowing the resonator network to handle previously unseen digit shapes from the test set. The modular structure provides a shape module which contains information about the object shape with translation factored out, allowing a simple classifier to operate on centered digits. The classification layer is trained solely on centered data, requiring much less training data, and the network as a whole can identify objects with arbitrary translations without data augmentation. The natural attention-like mechanism of the resonator network also allows for analysis of scenes with multiple objects, where the network dynamics selects and centers only one object at a time. Further, the specific position information of a particular object can be extracted from the translation module, and we show that the resonator can be designed to track multiple moving objects with precision of a few pixels.

We present OCRA, an Object-Centric framework for video-based human-to-Robot Action transfer that learns directly from human demonstration videos to enable robust manipulation. Object-centric learning emphasizes task-relevant objects and their interactions while filtering out irrelevant background, providing a natural and scalable way to teach robots. OCRA leverages multi-view RGB videos, the state-of-the-art 3D foundation model VGGT, and advanced detection and segmentation models to reconstruct object-centric 3D point clouds, capturing rich interactions between objects. To handle properties not easily perceived by vision alone, we incorporate tactile priors via a large-scale dataset of over one million tactile images. These 3D and tactile priors are fused through a multimodal module (ResFiLM) and fed into a Diffusion Policy to generate robust manipulation actions. Extensive experiments on both vision-only and visuo-tactile tasks show that OCRA significantly outperforms existing baselines and ablations, demonstrating its effectiveness for learning from human demonstration videos.

Current 4D representations decouple geometry, motion, and semantics: reconstruction methods discard interpretable motion structure; language-grounded methods attach semantics after motion is learned, blind to how objects move; and motion-aware methods encode dynamics as opaque per-point residuals without object-level organization. We propose 4D Synchronized Fields, a 4D Gaussian representation that learns object-factored motion in-loop during reconstruction and synchronizes language to the resulting kinematics through a per-object conditioned field. Each Gaussian trajectory is decomposed into shared object motion plus an implicit residual, and a kinematic-conditioned ridge map predicts temporal semantic variation, yielding a single representation in which reconstruction, motion, and semantics are structurally coupled and enabling open-vocabulary temporal queries that retrieve both objects and moments. On HyperNeRF, 4D Synchronized Fields achieves 28.52 dB mean PSNR, the highest among all language-grounded and motion-aware baselines, within 1.5 dB of reconstruction-only methods. On targeted temporal-state retrieval, the kinematic-conditioned field attains 0.884 mean accuracy, 0.815 mean vIoU, and 0.733 mean tIoU, surpassing 4D LangSplat (0.620, 0.433, and 0.439 respectively) and LangSplat (0.415, 0.304, and 0.262). Ablation confirms that kinematic conditioning is the primary driver, accounting for +0.45 tIoU over a static-embedding-only baseline. 4D Synchronized Fields is the only method that jointly exposes interpretable motion primitives and temporally grounded language fields from a single trained representation. Code will be released.

Slot attention has emerged as a powerful framework for unsupervised object-centric learning, decomposing visual scenes into a small set of compact vector representations called slots, each capturing a distinct region or object. However, these slots are learned in Euclidean space, which provides no geometric inductive bias for the hierarchical relationships that naturally structure visual scenes. In this work, we propose a simple post-hoc pipeline to project Euclidean slot embeddings onto the Lorentz hyperboloid of hyperbolic space, without modifying the underlying training pipeline. We construct five-level visual hierarchies directly from slot attention masks and analyse whether hyperbolic geometry reveals latent hierarchical structure that remains invisible in Euclidean space. Integrating our pipeline with SPOT (images), VideoSAUR (video), and SlotContrast (video), We find that hyperbolic projection exposes a consistent scene-level to object-level organisation, where coarse slots occupy greater manifold depth than fine slots, which is absent in Euclidean space. We further identify a "curvature--task tradeoff": low curvature ($c{=}0.2$) matches or outperforms Euclidean on parent slot retrieval, while moderate curvature ($c{=}0.5$) achieves better inter-level separation. Together, these findings suggest that slot representations already encode latent hierarchy that hyperbolic geometry reveals, motivating end-to-end hyperbolic training as a natural next step. Code and models are available at github.com/NeeluMadan/HHS.

Vision foundation models trained with self-supervised objectives achieve strong performance across diverse tasks and exhibit emergent object segmentation properties. However, their alignment with human object perception remains poorly understood. Here, we introduce a behavioral benchmark in which participants make same/different object judgments for dot pairs on naturalistic scenes, scaling up a classical psychophysics paradigm to over 1000 trials. We test a diverse set of vision models using a simple readout from their representations to predict subjects' reaction times. We observe a steady improvement across model generations, with both architecture and training objective contributing to alignment, and transformer-based models trained with the DINO self-supervised objective showing the strongest performance. To investigate the source of this improvement, we propose a novel metric to quantify the object-centric component of representations by measuring patch similarity within and between objects. Across models, stronger object-centric structure predicts human segmentation behavior more accurately. We further show that matching the Gram matrix of supervised transformer models, capturing similarity structure across image patches, with that of a self-supervised model through distillation improves their alignment with human behavior, converging with the prior finding that Gram anchoring improves DINOv3's feature quality. Together, these results demonstrate that self-supervised vision models capture object structure in a behaviorally human-like manner, and that Gram matrix structure plays a role in driving perceptual alignment.

Humans develop visual intelligence through perceiving and interacting with their environment - a self-supervised learning process grounded in egocentric experience. Inspired by this, we ask how can artificial systems learn stable object representations from continuous, uncurated first-person videos without relying on manual annotations. This setting poses challenges of separating, recognizing, and persistently tracking objects amid clutter, occlusion, and ego-motion. We propose EgoViT, a unified vision Transformer framework designed to learn stable object representations from unlabeled egocentric video. EgoViT bootstraps this learning process by jointly discovering and stabilizing "proto-objects" through three synergistic mechanisms: (1) Proto-object Learning, which uses intra-frame distillation to form discriminative representations; (2) Depth Regularization, which grounds these representations in geometric structure; and (3) Teacher-Filtered Temporal Consistency, which enforces identity over time. This creates a virtuous cycle where initial object hypotheses are progressively refined into stable, persistent representations. The framework is trained end-to-end on unlabeled first-person videos and exhibits robustness to geometric priors of varied origin and quality. On standard benchmarks, EgoViT achieves +8.0% CorLoc improvement in unsupervised object discovery and +4.8% mIoU improvement in semantic segmentation, demonstrating its potential to lay a foundation for robust visual abstraction in embodied intelligence.

For robotic agents operating in dynamic environments, learning visual state representations from streaming video observations is essential for sequential decision making. Recent self-supervised learning methods have shown strong transferability across vision tasks, but they do not explicitly address what a good visual state should encode. We argue that effective visual states must capture what-is-where by jointly encoding the semantic identities of scene elements and their spatial locations, enabling reliable detection of subtle dynamics across observations. To this end, we propose CroBo, a visual state representation learning framework based on a global-to-local reconstruction objective. Given a reference observation compressed into a compact bottleneck token, CroBo learns to reconstruct heavily masked patches in a local target crop from sparse visible cues, using the global bottleneck token as context. This learning objective encourages the bottleneck token to encode a fine-grained representation of scene-wide semantic entities, including their identities, spatial locations, and configurations. As a result, the learned visual states reveal how scene elements move and interact over time, supporting sequential decision making. We evaluate CroBo on diverse vision-based robot policy learning benchmarks, where it achieves state-of-the-art performance. Reconstruction analyses and perceptual straightness experiments further show that the learned representations preserve pixel-level scene composition and encode what-moves-where across observations. Project page available at: https://seokminlee-chris.github.io/CroBo-ProjectPage.

Slot-based object-centric learning represents an image as a set of latent slots with a decoder that combines them into an image or features. The decoder specifies how slots are combined into an output, but the slot set is typically fixed: the number of slots is chosen upfront and slots are only refined. This can lead to multiple slots competing for overlapping regions of the same entity rather than focusing on distinct regions. We introduce slot merging: a drop-in, lightweight operation on the slot set that merges overlapping slots during training. We quantify overlap with a Soft-IoU score between slot-attention maps and combine selected pairs via a barycentric update that preserves gradient flow. Merging follows a fixed policy, with the decision threshold inferred from overlap statistics, requiring no additional learnable modules. Integrated into the established feature-reconstruction pipeline of DINOSAUR, the proposed method improves object factorization and mask quality, surpassing other adaptive methods in object discovery and segmentation benchmarks.

Self-supervised learning (SSL) has made rapid progress, yet learned features often over-rely on contextual shortcuts-background textures and co-occurrence statistics. While video provides rich temporal variation, dense in-the-wild streams with strong ego-motion create a co-occurrence trap: foreground objects and background context move coherently, encouraging representations to collapse into scene encoders. To address this, we propose VINO (Video-driven Invariance for Non-Contextual Objects), a teacher-student framework that learns robust image encoders from dense video by imposing a structural information bottleneck. Using a class-agnostic structural prior solely to generate views-not as semantic pseudo-labels-VINO forms an asymmetric distillation problem. The teacher predicts from a foreground-union view with the background suppressed, while the student observes object-conditioned scene views that retain surrounding context but remove competing instances. Matching these targets via masked distillation makes background cues unreliable, pushing the representation toward object-centric invariances. We further enforce temporal object permanence via teacher-anchored cross-time distillation over track-matched objects, and stabilize part-to-whole consistency with mask-guided local views. Through attention visualization and unsupervised object discovery on PASCAL VOC, we demonstrate that VINO effectively disentangles foreground from background. Pretrained on the dense Walking Tours Venice video, VINO achieves 34.8 CorLoc, yielding highly focused, shape-biased representations that substantially outperform prior dense-video and motion-guided SSL baselines.

We introduce Latent Particle World Model (LPWM), a self-supervised object-centric world model scaled to real-world multi-object datasets and applicable in decision-making. LPWM autonomously discovers keypoints, bounding boxes, and object masks directly from video data, enabling it to learn rich scene decompositions without supervision. Our architecture is trained end-to-end purely from videos and supports flexible conditioning on actions, language, and image goals. LPWM models stochastic particle dynamics via a novel latent action module and achieves state-of-the-art results on diverse real-world and synthetic datasets. Beyond stochastic video modeling, LPWM is readily applicable to decision-making, including goal-conditioned imitation learning, as we demonstrate in the paper. Code, data, pre-trained models and video rollouts are available: https://taldatech.github.io/lpwm-web

Although data generation is often straightforward, extracting information from data is more difficult. Object-centric representation learning can extract information from images in an unsupervised manner. It does so by segmenting an image into its subcomponents: the objects. Each object is then represented in a low-dimensional latent space that can be used for downstream processing. Object-centric representation learning is dominated by autoencoder architectures (AEs). Here, we present ORGAN, a novel approach for object-centric representation learning, which is based on cycle-consistent Generative Adversarial Networks instead. We show that it performs similarly to other state-of-the-art approaches on synthetic datasets, while at the same time being the only approach tested here capable of handling more challenging real-world datasets with many objects and low visual contrast. Complementing these results, ORGAN creates expressive latent space representations that allow for object manipulation. Finally, we show that ORGAN scales well both with respect to the number of objects and the size of the images, giving it a unique edge over current state-of-the-art approaches.

Source-Free Domain Adaptive Object Detection (SF-DAOD) aims to adapt a detector trained on a labeled source domain to an unlabeled target domain without retaining any source data. Despite recent progress, most popular approaches focus on tuning pseudo-label thresholds or refining the teacher-student framework, while overlooking object-level structural cues within cross-domain data. In this work, we present CGSA, the first framework that brings Object-Centric Learning (OCL) into SF-DAOD by integrating slot-aware adaptation into the DETR-based detector. Specifically, our approach integrates a Hierarchical Slot Awareness (HSA) module into the detector to progressively disentangle images into slot representations that act as visual priors. These slots are then guided toward class semantics via a Class-Guided Slot Contrast (CGSC) module, maintaining semantic consistency and prompting domain-invariant adaptation. Extensive experiments on multiple cross-domain datasets demonstrate that our approach outperforms previous SF-DAOD methods, with theoretical derivations and experimental analysis further demonstrating the effectiveness of the proposed components and the framework, thereby indicating the promise of object-centric design in privacy-sensitive adaptation scenarios. Code is released at https://github.com/Michael-McQueen/CGSA.

Discrete image tokenizers provide a sequential interface for vision and multimodal models, but are typically optimized for reconstruction or compression and therefore tend to encode local appearance rather than object-level structure. We introduce COMiT, a communication-inspired framework for learning structured discrete visual representations. COMiT constructs a fixed-length latent message through sequential visual observations: at each step, a transformer processes a localized image crop and updates, refines, and reorganizes the existing token sequence. After several iterations, the resulting message conditions a flow-matching decoder that reconstructs the complete image. The encoder and decoder are implemented within a single transformer and trained end-to-end using flow-matching reconstruction and semantic representation-alignment objectives. COMiT substantially improves compositional generalization and relational reasoning over prior methods. Our experiments show that, while semantic alignment helps ground the representation, attentive sequential tokenization is critical for inducing more interpretable, object-centric token structures.

Compositional generalization, the ability to reason about novel combinations of familiar concepts, is fundamental to human cognition and a critical challenge for machine learning. Object-centric (OC) representations, which encode a scene as a set of objects, are often argued to support such generalization, but systematic evidence in visually rich settings is limited. We introduce a Visual Question Answering benchmark across three controlled visual worlds (CLEVRTex, Super-CLEVR, and MOVi-C) to measure how well vision encoders, with and without object-centric biases, generalize to unseen combinations of object properties. To ensure a fair and comprehensive comparison, we carefully account for training data diversity, sample size, representation size, downstream model capacity, and compute. We use DINOv2 and SigLIP2, two widely used vision encoders, as the foundation models and their OC counterparts. Our key findings reveal that (1) OC approaches are superior in harder compositional generalization settings; (2) original dense representations surpass OC only on easier settings and typically require substantially more downstream compute; and (3) OC models are more sample efficient, achieving stronger generalization with fewer images, whereas dense encoders catch up or surpass them only with sufficient data and diversity. Overall, object-centric representations offer stronger compositional generalization when any one of dataset size, training data diversity, or downstream compute is constrained.

Humans appear to represent objects when reasoning about physics with coarse, volumetric "bodies" that smooth concavities, trading fine visual detail for efficient physical predictions. Yet, the structure of these representations remains largely unknown. Segmentation models, in contrast, are trained for pixel-accurate masks that may misalign with such bodies. We ask whether and when these models nonetheless acquire human-like object representations. Using a time-to-collision (TTC) and change detection (CD) behavioral task with data from 178 and 50 human participants, respectively, we introduce a pipeline and an alignment metric to compare the visual representations of segmentation models to those of humans. We do this systematically on multiple architectures (DINOv2, SegFormer, DeepLabV3+, and UPerNet), varying their size and training time. We find that briefly trained models segment objects too coarsely, aligning poorly with humans, while fully trained models segment objects too finely. For each model, there is an intermediate training regime that best matches the coarse bodies observed in human behaviour, and larger models tend to reach it earlier. We show these bodies emerge under resource constraints in general-purpose vision models, providing computational support to resource-rational accounts of human cognition. This work provides a foundational framework for testing alignment between vision models and humans and shows there is a growing gap between the state-of-the-art in artificial intelligence and human cognition, driven by scaling model size and training.

The generalization capabilities of robotic manipulation policies are heavily influenced by the choice of visual representations. Existing approaches typically rely on representations extracted from pre-trained encoders, using two dominant types of features: global features, which summarize an entire image via a single pooled vector, and dense features, which preserve a patch-wise embedding from the final encoder layer. While widely used, both feature types mix task-relevant and irrelevant information, leading to poor generalization under distribution shifts, such as changes in lighting, textures, or the presence of distractors. In this work, we explore an intermediate structured alternative: Slot-Based Object-Centric Representations (SBOCR), which group dense features into a finite set of object-like entities. This representation permits to naturally reduce the noise provided to the robotic manipulation policy while keeping enough information to efficiently perform the task. We benchmark a range of global and dense representations against intermediate slot-based representations, across a suite of simulated and real-world manipulation tasks ranging from simple to complex. We evaluate their generalization under diverse visual conditions, including changes in lighting, texture, and the presence of distractors. Our findings reveal that SBOCR-based policies outperform dense and global representation-based policies in generalization settings, even without task-specific pretraining. These insights suggest that SBOCR is a promising direction for designing visual systems that generalize effectively in dynamic, real-world robotic environments.

Visual foundation models provide strong perceptual features for robotics, but their dense representations lack explicit object-level structure, limiting robustness and controllability in manipulation tasks. We propose STORM (Slot-based Task-aware Object-centric Representation for robotic Manipulation), a lightweight object-centric adaptation module that augments frozen visual foundation models with a small set of task-aware slots for robotic manipulation. Rather than fully tuning large backbones on the task, STORM employs an efficient two-stage training strategy: few layers of object-centric representation are first trained on top of the frozen backbone through visual--semantic pretraining using language embeddings, then jointly adapted with a downstream manipulation policy for task alignement. This staged learning prevents degenerate slot formation and preserves semantic consistency while aligning perception with task objectives. Experiments on object discovery benchmarks and robotic manipulation tasks show that STORM improves control performance and generalization to visual shifts (distractors, textures, lighting) compared to directly using frozen or fine-tuned foundation model features, or existing object-centric representations. STORM serves not only as an efficient mechanism for refining generic foundation model features, but also as a novel way of injecting beneficial structural and semantic bias into policy learning.