Simple 3D Pose Features Support Human and Machine Social Scene Understanding

• URL: http://arxiv.org/abs/2511.03988v1
• Date: Thu, 06 Nov 2025 02:19:26 GMT
• Title: Simple 3D Pose Features Support Human and Machine Social Scene Understanding
• Authors: Wenshuo Qin, Leyla Isik,
• Abstract summary: We show that humans rely on 3D visuospatial pose information to make social interaction judgments.<n>We extract 3D joint positions of people in short video clips depicting everyday human actions.<n>Strikingly, 3D joint positions outperformed most current AI vision models.
• Score: 1.411894456054802
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Humans can quickly and effortlessly extract a variety of information about others' social interactions from visual input, ranging from visuospatial cues like whether two people are facing each other to higher-level information. Yet, the computations supporting these abilities remain poorly understood, and social interaction recognition continues to challenge even the most advanced AI vision systems. Here, we hypothesized that humans rely on 3D visuospatial pose information to make social interaction judgments, which is absent in most AI vision models. To test this, we combined state-of-the-art pose and depth estimation algorithms to extract 3D joint positions of people in short video clips depicting everyday human actions and compared their ability to predict human social interaction judgments with current AI vision models. Strikingly, 3D joint positions outperformed most current AI vision models, revealing that key social information is available in explicit body position but not in the learned features of most vision models, including even the layer-wise embeddings of the pose models used to extract joint positions. To uncover the critical pose features humans use to make social judgments, we derived a compact set of 3D social pose features describing only the 3D position and direction of faces in the videos. We found that these minimal descriptors matched the predictive strength of the full set of 3D joints and significantly improved the performance of off-the-shelf AI vision models when combined with their embeddings. Moreover, the degree to which 3D social pose features were represented in each off-the-shelf AI vision model predicted the model's ability to match human social judgments. Together, our findings provide strong evidence that human social scene understanding relies on explicit representations of 3D pose and can be supported by simple, structured visuospatial primitives.

Related papers

• Human-level 3D shape perception emerges from multi-view learning [63.048728487674815]
  We develop a modeling framework that predicts human 3D shape inferences for arbitrary objects.<n>We achieve this with a novel class of neural networks trained using a visual-spatial objective over naturalistic sensory data.<n>We find that human-level 3D perception can emerge from a simple, scalable learning objective over naturalistic visual-spatial data.
  arXiv  Detail & Related papers  (2026-02-19T18:56:05Z)
• Social 3D Scene Graphs: Modeling Human Actions and Relations for Interactive Service Robots [5.8503433899583905]
  We introduce Social 3D Scene Graphs, an augmented 3D Scene Graph representation that captures humans, their attributes, activities and relationships in the environment, both local and remote.<n>Our representation improves human activity prediction and reasoning about human-environment relations, paving the way toward socially intelligent robots.
  arXiv  Detail & Related papers  (2025-09-29T16:00:40Z)
• Zero-Shot Human-Object Interaction Synthesis with Multimodal Priors [31.277540988829976]
  This paper proposes a novel zero-shot HOI synthesis framework without relying on end-to-end training on currently limited 3D HOI datasets.<n>We employ pre-trained human pose estimation models to extract human poses and introduce a generalizable category-level 6-DoF estimation method to obtain the object poses from 2D HOI images.
  arXiv  Detail & Related papers  (2025-03-25T23:55:47Z)
• Beyond the Contact: Discovering Comprehensive Affordance for 3D Objects from Pre-trained 2D Diffusion Models [8.933560282929726]
  We introduce a novel affordance representation, named Comprehensive Affordance (ComA) Given a 3D object mesh, ComA models the distribution of relative orientation and proximity of vertices in interacting human meshes. We demonstrate that ComA outperforms competitors that rely on human annotations in modeling contact-based affordance.
  arXiv  Detail & Related papers  (2024-01-23T18:59:59Z)
• Social-Transmotion: Promptable Human Trajectory Prediction [65.80068316170613]
  Social-Transmotion is a generic Transformer-based model that exploits diverse and numerous visual cues to predict human behavior.<n>Our approach is validated on multiple datasets, including JTA, JRDB, Pedestrians and Cyclists in Road Traffic, and ETH-UCY.
  arXiv  Detail & Related papers  (2023-12-26T18:56:49Z)
• Probabilistic Human Mesh Recovery in 3D Scenes from Egocentric Views [32.940614931864154]
  We propose a scene-conditioned diffusion method to model the body pose distribution. The method generates bodies in plausible human-scene interactions. It achieves superior accuracy for visible joints and diversity for invisible body parts.
  arXiv  Detail & Related papers  (2023-04-12T17:58:57Z)
• Neural Novel Actor: Learning a Generalized Animatable Neural Representation for Human Actors [98.24047528960406]
  We propose a new method for learning a generalized animatable neural representation from a sparse set of multi-view imagery of multiple persons. The learned representation can be used to synthesize novel view images of an arbitrary person from a sparse set of cameras, and further animate them with the user's pose control.
  arXiv  Detail & Related papers  (2022-08-25T07:36:46Z)
• Unsupervised Cross-Modal Alignment for Multi-Person 3D Pose Estimation [52.94078950641959]
  We present a deployment friendly, fast bottom-up framework for multi-person 3D human pose estimation. We adopt a novel neural representation of multi-person 3D pose which unifies the position of person instances with their corresponding 3D pose representation. We propose a practical deployment paradigm where paired 2D or 3D pose annotations are unavailable.
  arXiv  Detail & Related papers  (2020-08-04T07:54:25Z)
• HMOR: Hierarchical Multi-Person Ordinal Relations for Monocular Multi-Person 3D Pose Estimation [54.23770284299979]
  This paper introduces a novel form of supervision - Hierarchical Multi-person Ordinal Relations (HMOR) HMOR encodes interaction information as the ordinal relations of depths and angles hierarchically. An integrated top-down model is designed to leverage these ordinal relations in the learning process. The proposed method significantly outperforms state-of-the-art methods on publicly available multi-person 3D pose datasets.
  arXiv  Detail & Related papers  (2020-08-01T07:53:27Z)
• Self-Supervised 3D Human Pose Estimation via Part Guided Novel Image Synthesis [72.34794624243281]
  We propose a self-supervised learning framework to disentangle variations from unlabeled video frames. Our differentiable formalization, bridging the representation gap between the 3D pose and spatial part maps, allows us to operate on videos with diverse camera movements.
  arXiv  Detail & Related papers  (2020-04-09T07:55:01Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.