Shiry Ginosar

Evonne Ng, Sanjay Subramanian, Dan Klein et al. · berkeley

We present a framework for generating appropriate facial responses from a listener in dyadic social interactions based on the speaker's words. Given an input transcription of the speaker's words with their timestamps, our approach autoregressively predicts a response of a listener: a sequence of listener facial gestures, quantized using a VQ-VAE. Since gesture is a language component, we propose treating the quantized atomic motion elements as additional language token inputs to a transformer-based large language model. Initializing our transformer with the weights of a language model pre-trained only on text results in significantly higher quality listener responses than training a transformer from scratch. We show that our generated listener motion is fluent and reflective of language semantics through quantitative metrics and a qualitative user study. In our evaluation, we analyze the model's ability to utilize temporal and semantic aspects of spoken text. Project page: https://people.eecs.berkeley.edu/~evonne_ng/projects/text2listen/

A. Sophia Koepke, Daniil Zverev, Shiry Ginosar et al. · berkeley

The Platonic Representation Hypothesis suggests that neural networks trained on different modalities (e.g., text and images) align and eventually converge toward the same representation of reality. If true, this has significant implications for whether modality choice matters at all. We show that the experimental evidence for this hypothesis is fragile and depends critically on the evaluation regime. Alignment is measured using mutual nearest neighbors on small datasets ($\approx$1K samples) and degrades substantially as the dataset is scaled to millions of samples. The alignment that remains between model representations reflects coarse semantic overlap rather than consistent fine-grained structure. Moreover, the evaluations in Huh et al. are done in a one-to-one image-caption setting, a constraint that breaks down in realistic many-to-many settings and further reduces alignment. We also find that the reported trend of stronger language models increasingly aligning with vision does not appear to hold for newer models. Overall, our findings suggest that the current evidence for cross-modal representational convergence is considerably weaker than subsequent works have taken it to be. Models trained on different modalities may learn equally rich representations of the world, just not the same one.

Ioannis Siglidis, Aleksander Holynski, Alexei A. Efros et al. · berkeley

This paper demonstrates how to use generative models trained for image synthesis as tools for visual data mining. Our insight is that since contemporary generative models learn an accurate representation of their training data, we can use them to summarize the data by mining for visual patterns. Concretely, we show that after finetuning conditional diffusion models to synthesize images from a specific dataset, we can use these models to define a typicality measure on that dataset. This measure assesses how typical visual elements are for different data labels, such as geographic location, time stamps, semantic labels, or even the presence of a disease. This analysis-by-synthesis approach to data mining has two key advantages. First, it scales much better than traditional correspondence-based approaches since it does not require explicitly comparing all pairs of visual elements. Second, while most previous works on visual data mining focus on a single dataset, our approach works on diverse datasets in terms of content and scale, including a historical car dataset, a historical face dataset, a large worldwide street-view dataset, and an even larger scene dataset. Furthermore, our approach allows for translating visual elements across class labels and analyzing consistent changes.

Evonne Ng, Hanbyul Joo, Liwen Hu et al.

We present a framework for modeling interactional communication in dyadic conversations: given multimodal inputs of a speaker, we autoregressively output multiple possibilities of corresponding listener motion. We combine the motion and speech audio of the speaker using a motion-audio cross attention transformer. Furthermore, we enable non-deterministic prediction by learning a discrete latent representation of realistic listener motion with a novel motion-encoding VQ-VAE. Our method organically captures the multimodal and non-deterministic nature of nonverbal dyadic interactions. Moreover, it produces realistic 3D listener facial motion synchronous with the speaker (see video). We demonstrate that our method outperforms baselines qualitatively and quantitatively via a rich suite of experiments. To facilitate this line of research, we introduce a novel and large in-the-wild dataset of dyadic conversations. Code, data, and videos available at https://evonneng.github.io/learning2listen/.

Eunice Yiu, Maan Qraitem, Anisa Noor Majhi et al. · berkeley

This paper investigates visual analogical reasoning in large multimodal models (LMMs) compared to human adults and children. A "visual analogy" is an abstract rule inferred from one image and applied to another. While benchmarks exist for testing visual reasoning in LMMs, they require advanced skills and omit basic visual analogies that even young children can make. Inspired by developmental psychology, we propose a new benchmark of 4,300 visual transformations of everyday objects to test LMMs on visual analogical reasoning and compare them to children (ages three to five) and to adults. We structure the evaluation into three stages: identifying what changed (e.g., color, number, etc.), how it changed (e.g., added one object), and applying the rule to new scenarios. Our findings show that while GPT-o1, GPT-4V, LLaVA-1.5, and MANTIS identify the "what" effectively, they struggle with quantifying the "how" and extrapolating this rule to new objects. In contrast, children and adults exhibit much stronger analogical reasoning at all three stages. Additionally, the strongest tested model, GPT-o1, performs better in tasks involving simple surface-level visual attributes like color and size, correlating with quicker human adult response times. Conversely, more complex tasks such as number, rotation, and reflection, which necessitate extensive cognitive processing and understanding of extrinsic spatial properties in the physical world, present more significant challenges. Altogether, these findings highlight the limitations of training models on data that primarily consists of 2D images and text.

Vongani Maluleke, Lea Müller, Jathushan Rajasegaran et al.

This paper asks to what extent social interaction influences one's behavior. We study this in the setting of two dancers dancing as a couple. We first consider a baseline in which we predict a dancer's future moves conditioned only on their past motion without regard to their partner. We then investigate the advantage of taking social information into account by conditioning also on the motion of their dancing partner. We focus our analysis on Swing, a dance genre with tight physical coupling for which we present an in-the-wild video dataset. We demonstrate that single-person future motion prediction in this context is challenging. Instead, we observe that prediction greatly benefits from considering the interaction partners' behavior, resulting in surprisingly compelling couple dance synthesis results (see supp. video). Our contributions are a demonstration of the advantages of socially conditioned future motion prediction and an in-the-wild, couple dance video dataset to enable future research in this direction. Video results are available on the project website: https://von31.github.io/synNsync

Eunice Yiu, Kelsey Allen, Shiry Ginosar et al.

Learning about the causal structure of the world is a fundamental problem for human cognition. Causal models and especially causal learning have proved to be difficult for large pretrained models using standard techniques of deep learning. In contrast, cognitive scientists have applied advances in our formal understanding of causation in computer science, particularly within the Causal Bayes Net formalism, to understand human causal learning. In the very different tradition of reinforcement learning, researchers have described an intrinsic reward signal called "empowerment" which maximizes mutual information between actions and their outcomes. "Empowerment" may be an important bridge between classical Bayesian causal learning and reinforcement learning and may help to characterize causal learning in humans and enable it in machines. If an agent learns an accurate causal world model, they will necessarily increase their empowerment, and increasing empowerment will lead to a more accurate causal world model. Empowerment may also explain distinctive features of childrens causal learning, as well as providing a more tractable computational account of how that learning is possible. In an empirical study, we systematically test how children and adults use cues to empowerment to infer causal relations, and design effective causal interventions.

Sanjay Subramanian, Evonne Ng, Lea Müller et al.

Language is often used to describe physical interaction, yet most 3D human pose estimation methods overlook this rich source of information. We bridge this gap by leveraging large multimodal models (LMMs) as priors for reconstructing contact poses, offering a scalable alternative to traditional methods that rely on human annotations or motion capture data. Our approach extracts contact-relevant descriptors from an LMM and translates them into tractable losses to constrain 3D human pose optimization. Despite its simplicity, our method produces compelling reconstructions for both two-person interactions and self-contact scenarios, accurately capturing the semantics of physical and social interactions. Our results demonstrate that LMMs can serve as powerful tools for contact prediction and pose estimation, offering an alternative to costly manual human annotations or motion capture data. Our code is publicly available at https://prosepose.github.io.

Caroline Chan, Shiry Ginosar, Tinghui Zhou et al.

This paper presents a simple method for "do as I do" motion transfer: given a source video of a person dancing, we can transfer that performance to a novel (amateur) target after only a few minutes of the target subject performing standard moves. We approach this problem as video-to-video translation using pose as an intermediate representation. To transfer the motion, we extract poses from the source subject and apply the learned pose-to-appearance mapping to generate the target subject. We predict two consecutive frames for temporally coherent video results and introduce a separate pipeline for realistic face synthesis. Although our method is quite simple, it produces surprisingly compelling results (see video). This motivates us to also provide a forensics tool for reliable synthetic content detection, which is able to distinguish videos synthesized by our system from real data. In addition, we release a first-of-its-kind open-source dataset of videos that can be legally used for training and motion transfer.

Neerja Thakkar, Shiry Ginosar, Jacob Walker et al.

Visual intelligence requires anticipating the future behavior of agents, yet vision systems lack a general representation for motion and behavior. We propose dense point trajectories as visual tokens for behavior, a structured mid-level representation that disentangles motion from appearance and generalizes across diverse non-rigid agents, such as animals in-the-wild. Building on this abstraction, we design a diffusion transformer that models unordered sets of trajectories and explicitly reasons about occlusion, enabling coherent forecasts of complex motion patterns. To evaluate at scale, we curate 300 hours of unconstrained animal video with robust shot detection and camera-motion compensation. Experiments show that forecasting trajectory tokens achieves category-agnostic, data-efficient prediction, outperforms state-of-the-art baselines, and generalizes to rare species and morphologies, providing a foundation for predictive visual intelligence in the wild.

Jathushan Rajasegaran, Xinlei Chen, Rulilong Li et al.

This paper explores Masked Autoencoders (MAE) with Gaussian Splatting. While reconstructive self-supervised learning frameworks such as MAE learns good semantic abstractions, it is not trained for explicit spatial awareness. Our approach, named Gaussian Masked Autoencoder, or GMAE, aims to learn semantic abstractions and spatial understanding jointly. Like MAE, it reconstructs the image end-to-end in the pixel space, but beyond MAE, it also introduces an intermediate, 3D Gaussian-based representation and renders images via splatting. We show that GMAE can enable various zero-shot learning capabilities of spatial understanding (e.g., figure-ground segmentation, image layering, edge detection, etc.) while preserving the high-level semantics of self-supervised representation quality from MAE. To our knowledge, we are the first to employ Gaussian primitives in an image representation learning framework beyond optimization-based single-scene reconstructions. We believe GMAE will inspire further research in this direction and contribute to developing next-generation techniques for modeling high-fidelity visual data. More details at https://brjathu.github.io/gmae

Neerja Thakkar, Tara Sadjadpour, Jathushan Rajasegaran et al.

We introduce a simple framework for predicting the behavior of an agent in multi-agent settings. In contrast to autoregressive (AR) tasks, such as language processing, our focus is on scenarios with multiple agents whose interactions are shaped by physical constraints and internal motivations. To this end, we propose Poly-Autoregressive (PAR) modeling, which forecasts an ego agent's future behavior by reasoning about the ego agent's state history and the past and current states of other interacting agents. At its core, PAR represents the behavior of all agents as a sequence of tokens, each representing an agent's state at a specific timestep. With minimal data pre-processing changes, we show that PAR can be applied to three different problems: human action forecasting in social situations, trajectory prediction for autonomous vehicles, and object pose forecasting during hand-object interaction. Using a small proof-of-concept transformer backbone, PAR outperforms AR across these three scenarios. The project website can be found at https://neerja.me/PAR/.

Jacob C Walker, Pedro Vélez, Luisa Polania Cabrera et al.

Forecasting what will happen next is a critical skill for general-purpose systems that plan or act in the world at different levels of abstraction. In this paper, we identify a strong correlation between a vision model's perceptual ability and its generalist forecasting performance over short time horizons. This trend holds across a diverse set of pretrained models-including those trained generatively-and across multiple levels of abstraction, from raw pixels to depth, point tracks, and object motion. The result is made possible by a novel generalist forecasting framework that operates on any frozen vision backbone: we train latent diffusion models to forecast future features in the frozen representation space, which are then decoded via lightweight, task-specific readouts. To enable consistent evaluation across tasks, we introduce distributional metrics that compare distributional properties directly in the space of downstream tasks and apply this framework to nine models and four tasks. Our results highlight the value of bridging representation learning and generative modeling for temporally grounded video understanding.

Medhini Narasimhan, Shiry Ginosar, Andrew Owens et al.

We introduce a non-parametric approach for infinite video texture synthesis using a representation learned via contrastive learning. We take inspiration from Video Textures, which showed that plausible new videos could be generated from a single one by stitching its frames together in a novel yet consistent order. This classic work, however, was constrained by its use of hand-designed distance metrics, limiting its use to simple, repetitive videos. We draw on recent techniques from self-supervised learning to learn this distance metric, allowing us to compare frames in a manner that scales to more challenging dynamics, and to condition on other data, such as audio. We learn representations for video frames and frame-to-frame transition probabilities by fitting a video-specific model trained using contrastive learning. To synthesize a texture, we randomly sample frames with high transition probabilities to generate diverse temporally smooth videos with novel sequences and transitions. The model naturally extends to an audio-conditioned setting without requiring any finetuning. Our model outperforms baselines on human perceptual scores, can handle a diverse range of input videos, and can combine semantic and audio-visual cues in order to synthesize videos that synchronize well with an audio signal.

Andrew Liu, Shiry Ginosar, Tinghui Zhou et al.

We propose a learning-based framework for disentangling outdoor scenes into temporally-varying illumination and permanent scene factors. Inspired by the classic intrinsic image decomposition, our learning signal builds upon two insights: 1) combining the disentangled factors should reconstruct the original image, and 2) the permanent factors should stay constant across multiple temporal samples of the same scene. To facilitate training, we assemble a city-scale dataset of outdoor timelapse imagery from Google Street View, where the same locations are captured repeatedly through time. This data represents an unprecedented scale of spatio-temporal outdoor imagery. We show that our learned disentangled factors can be used to manipulate novel images in realistic ways, such as changing lighting effects and scene geometry. Please visit factorize-a-city.github.io for animated results.

Evonne Ng, Shiry Ginosar, Trevor Darrell et al.

We propose a novel learned deep prior of body motion for 3D hand shape synthesis and estimation in the domain of conversational gestures. Our model builds upon the insight that body motion and hand gestures are strongly correlated in non-verbal communication settings. We formulate the learning of this prior as a prediction task of 3D hand shape over time given body motion input alone. Trained with 3D pose estimations obtained from a large-scale dataset of internet videos, our hand prediction model produces convincing 3D hand gestures given only the 3D motion of the speaker's arms as input. We demonstrate the efficacy of our method on hand gesture synthesis from body motion input, and as a strong body prior for single-view image-based 3D hand pose estimation. We demonstrate that our method outperforms previous state-of-the-art approaches and can generalize beyond the monologue-based training data to multi-person conversations. Video results are available at http://people.eecs.berkeley.edu/~evonne_ng/projects/body2hands/.

Shiry Ginosar, Amir Bar, Gefen Kohavi et al.

Human speech is often accompanied by hand and arm gestures. Given audio speech input, we generate plausible gestures to go along with the sound. Specifically, we perform cross-modal translation from "in-the-wild'' monologue speech of a single speaker to their hand and arm motion. We train on unlabeled videos for which we only have noisy pseudo ground truth from an automatic pose detection system. Our proposed model significantly outperforms baseline methods in a quantitative comparison. To support research toward obtaining a computational understanding of the relationship between gesture and speech, we release a large video dataset of person-specific gestures. The project website with video, code and data can be found at http://people.eecs.berkeley.edu/~shiry/speech2gesture .

L. Jason Anastasopoulos, Dhruvil Badani, Crystal Lee et al.

While members of Congress now routinely communicate with constituents using images on a variety of internet platforms, little is known about how images are used as a means of strategic political communication. This is due primarily to computational limitations which have prevented large-scale, systematic analyses of image features. New developments in computer vision, however, are bringing the systematic study of images within reach. Here, we develop a framework for understanding visual political communication by extending Fenno's analysis of home style (Fenno 1978) to images and introduce "photographic" home styles. Using approximately 192,000 photographs collected from MCs Facebook profiles, we build machine learning software with convolutional neural networks and conduct an image manipulation experiment to explore how the race of people that MCs pose with shape photographic home styles. We find evidence that electoral pressures shape photographic home styles and demonstrate that Democratic and Republican members of Congress use images in very different ways.

Shiry Ginosar, Kate Rakelly, Sarah Sachs et al.

Imagery offers a rich description of our world and communicates a volume and type of information that cannot be captured by text alone. Since the invention of the camera, an ever-increasing number of photographs document our "visual culture" complementing historical texts. But currently, this treasure trove of knowledge can only be analyzed manually by historians, and only at small scale. In this paper we perform automated analysis on a large-scale historical image dataset. Our main contributions are: 1) A publicly-available dataset of 168,055 (37,921 frontal-facing) American high school yearbook portraits. 2) Weakly-supervised data-driven techniques to discover historical visual trends in fashion and identify date-specific visual patterns. 3) A classifier to predict when a portrait was taken, with median error of 4 years for women and 6 for men. 4) A new method for discovering and displaying the visual elements used by the CNN-based date-prediction model to date portraits, finding that they correspond to the tell-tale fashions of each era. Project page can be found at: http://people.eecs.berkeley.edu/~shiry/projects/yearbooks/yearbooks.html .

Shiry Ginosar, Daniel Haas, Timothy Brown et al.

Although the human visual system is surprisingly robust to extreme distortion when recognizing objects, most evaluations of computer object detection methods focus only on robustness to natural form deformations such as people's pose changes. To determine whether algorithms truly mirror the flexibility of human vision, they must be compared against human vision at its limits. For example, in Cubist abstract art, painted objects are distorted by object fragmentation and part-reorganization, to the point that human vision often fails to recognize them. In this paper, we evaluate existing object detection methods on these abstract renditions of objects, comparing human annotators to four state-of-the-art object detectors on a corpus of Picasso paintings. Our results demonstrate that while human perception significantly outperforms current methods, human perception and part-based models exhibit a similarly graceful degradation in object detection performance as the objects become increasingly abstract and fragmented, corroborating the theory of part-based object representation in the brain.