NeurIPS 2025 Papers

We introduce MusclePose as an end-to-end learnable physics-infused 3D human pose estimator that incorporates muscle-dynamics modeling to infer human dynamics from monocular video. Current physics pose estimators aim to predict physically plausible poses by enforcing the underlying dynamics equations that govern motion. Since this is an underconstrained problem without force-annotated data, methods often estimate kinetics with external physics optimizers that may not be compatible with existing learning frameworks, or are too slow for real-time inference. While more recent methods use a regression-based approach to overcome these issues, the estimated kinetics can be seen as auxiliary predictions, and may not be physically plausible. To this end, we build on existing regression-based approaches, and aim to improve the biofidelity of kinetic inference with a multihypothesis approach --- by inferring joint torques via Lagrange’s equations and via muscle dynamics modeling with muscle torque generators. Furthermore, MusclePose predicts detailed human anthropometrics based on values from biomechanics studies, in contrast to existing physics pose estimators that construct their human models with shape primitives. We show that MusclePose is competitive with existing 3D pose estimators in positional accuracy, while also able to infer plausible human kinetics and muscle signals consistent with values from biomechanics studies, without requiring an external physics engine.

3D Human Pose Estimation (HPE) is a fundamental task in the computer vision. Generalization in 3D HPE task is crucial due to the need for robustness across diverse environments and datasets. Existing methods often focus on learning relationships between joints to enhance the generalization capability, but the role of the loss landscape, which is closely tied to generalization, remains underexplored. In this paper, we empirically visualize the loss landscape of the 3D HPE task, revealing its complexity and the challenges it poses for optimization. To address this, we first introduce a simple adaptive scaling mechanism that smooths the loss landscape. We further observe that different solutions on this smoothed loss landscape exhibit varying generalization behaviors. Based on this insight, we propose an efficient ensemble approach that combines diverse solutions on the smooth loss landscape induced by our adaptive scaling mechanism. Extensive experimental results demonstrate that our approach improves the generalization capability of 3D HPE models, and can be easily applied, regardless of model architecture, with consistent performance gains.

We study online preference-based reinforcement learning (PbRL) with the goal of improving sample efficiency. While a growing body of theoretical work has emerged—motivated by PbRL’s recent empirical success, particularly in aligning large language models (LLMs)—most existing studies focus only on pairwise comparisons. A few recent works (Zhu et al., 2023, Mukherjee et al., 2024, Thekumparampil et al., 2024) have explored using multiple comparisons and ranking feedback, but their performance guarantees fail to improve—and can even deteriorate—as the feedback length increases, despite the richer information available. To address this gap, we adopt the Plackett–Luce (PL) model for ranking feedback over action subsets and propose **M-AUPO**, an algorithm that selects multiple actions by maximizing the average uncertainty within the offered subset. We prove that **M-AUPO** achieves a suboptimality gap of $\tilde{\mathcal{O}}\left( \frac{d}{T} \sqrt{ \sum_{t=1}^T \frac{1}{|S_t|}} \right)$, where $T$ is the total number of rounds, $d$ is the feature dimension, and $|S_t|$ is the size of the subset at round $t$. This result shows that larger subsets directly lead to improved performance and, notably, the bound avoids the exponential dependence on the unknown parameter’s norm, which was a fundamental limitation in most previous works. Moreover, we establish a near-matching lower bound of $\Omega \left( \frac{d}{K \sqrt{T}} \right)$, where $K$ is the maximum subset size. To the best of our knowledge, this is the first theoretical result in PbRL with ranking feedback that explicitly shows improved sample efficiency as a function of the subset size.

Diffusion models have yielded remarkable success in virtual try-on (VTON) task, yet they often fall short of fully meeting user expectations regarding visual quality and detail preservation. To alleviate this issue, we curate a dataset of synthesized VTON images annotated with human judgments across multiple perceptual criteria. A vision large language model (VLLM), namely VTON-VLLM, is then learnt on these annotations. VTON-VLLM functions as a unified ``fashion expert'' and is capable of both evaluating and steering VTON synthesis towards human preferences. Technically, beyond serving as an automatic VTON evaluator, VTON-VLLM upgrades VTON model through two pivotal ways: (1) providing fine-grained supervisory signals during the training of a plug-and-play VTON refinement model, and (2) enabling adaptive and preference-aware test-time scaling at inference. To benchmark VTON models more holistically, we introduce VITON-Bench, a challenging test suite of complex try-on scenarios, and human-preference–aware metrics. Extensive experiments demonstrate that powering VTON models with our VTON-VLLM markedly enhances alignment with human preferences. Code is publicly available at: [https://github.com/HiDream-ai/VTON-VLLM/](https://github.com/HiDream-ai/VTON-VLLM/).

People's goal-directed behaviors are influenced by their cognitive biases, and autonomous systems that interact with people should be aware of this. For example, people's attention to objects in their environment will be biased in a way that systematically affects how they perform everyday tasks such as driving to work. Here, building on recent work in computational cognitive science, we formally articulate the \textit{attention-aware inverse planning problem}, in which the goal is to estimate a person's attentional biases from their actions. We demonstrate how attention-aware inverse planning systematically differs from standard inverse reinforcement learning and how cognitive biases can be inferred from behavior. Finally, we present an approach to attention-aware inverse planning that combines deep reinforcement learning with computational cognitive modeling. We use this approach to infer the attentional strategies of RL agents in real-life driving scenarios selected from the Waymo Open Dataset, demonstrating the scalability of estimating cognitive biases with attention-aware inverse planning.

The task of localizing an object's spatial tube based on language instructions and video, known as spatial video grounding (SVG), has attracted widespread interest. Existing SVG tasks have focused on ego-centric fixed front perspective and simple scenes, which only involved a very limited view and environment. However, UAV-based SVG remains underexplored, which neglects the inherent disparities in drone movement and the complexity of aerial object localization. To facilitate research in this field, we introduce the novel spatial aerial video grounding (SAVG) task. Specifically, we meticulously construct a large-scale benchmark, UAV-SVG, which contains over 2 million frames and offers 216 highly diverse target categories. To address the disparities and challenges posed by complex aerial environments, we propose a new end-to-end transformer architecture, coined SAVG-DETR. The innovations are three-fold. 1) To overcome the computational explosion of self-attention when introducing multi-scale features, our encoder efficiently decouples the multi-modality and multi-scale spatio-temporal modeling into intra-scale multi-modality interaction and cross-scale visual-only fusion. 2) To enhance small object grounding ability, we propose the language modulation module to integrate multi-scale information into language features and the multi-level progressive spatial decoder to decode from high to low level. The decoding stage for the lower-level vision-language features is gradually increased. 3) To improve the prediction consistency across frames, we design the decoding paradigm based on offset generation. At each decoding stage, we utilize reference anchors to constrict the grounding region, use context-rich object queries to predict offsets, and update reference anchors for the next stage. From coarse to fine, our SAVG-DETR gradually bridges the modality gap and iteratively refines reference anchors of the referred object, eventually grounding the spatial tube. Extensive experiments demonstrate that our SAVG-DETR significantly outperforms existing state-of-the-art methods. The dataset and code will be available at here.

Learning robot control policies from human videos is a promising direction for scaling up robot learning. However, how to extract action knowledge (or action representations) from videos for policy learning remains a key challenge. Existing action representations such as video frames, pixelflow, and pointcloud flow have inherent limitations such as modeling complexity or loss of information. In this paper, we propose to use object-centric 3D motion field to represent actions for robot learning from human videos, and present a novel framework for extracting this representation from videos for zero-shot control. We introduce two novel components. First, a novel training pipeline for training a ``denoising'' 3D motion field estimator to extract fine object 3D motions from human videos with noisy depth robustly. Second, a dense object-centric 3D motion field prediction architecture that favors both cross-embodiment transfer and policy generalization to background. We evaluate the system in real world setups. Experiments show that our method reduces 3D motion estimation error by over 50% compared to the latest method, achieve 55% average success rate in diverse tasks where prior approaches fail ($\lesssim 10$\%), and can even acquire fine-grained manipulation skills like insertion.

This work tackles a challenging problem: stochastic human motion prediction (SHMP), which aims to forecast diverse and physically plausible future pose sequences based on a short history of observed motion. While autoregressive sequence models have excelled in related generation tasks, their reliance on vector‐quantized tokenization limits motion fidelity and training stability. To overcome these drawbacks, we introduce \textbf{WaveAR}, a novel AR based framework which is the first successful application of a continuous autoregressive generation paradigm to HMP to our best knowledge. WaveAR consists of two stages. In the first stage, a lightweight Spatio‐Temporal VAE (ST-VAE) compresses the raw 3D-joint sequence into a downsampled latent token stream, providing a compact yet expressive foundation. In the second stage, we apply masked autoregressive prediction directly in this continuous latent space, conditioning on both unmasked latents and multi‐scale spectral cues extracted via a 2D discrete wavelet transform. A fusion module consisting of alternating cross-attention and self-attention layers adaptively fuses temporal context with low- and high-frequency wavelet subbands, and a small MLP‐based diffusion head predicts per-token noise residuals under a denoising loss. By avoiding vector quantization and integrating localized frequency information, WaveAR preserves fine‐grained motion details while maintaining fast inference speed. Extensive experiments on standard benchmarks demonstrate that our approach delivers more accurate and computationally efficient predictions than prior state‐of-the-art methods.

From media platforms to chatbots, algorithms shape how people interact, learn, and discover information. Such interactions between users and an algorithm often unfold over multiple steps, during which strategic users can guide the algorithm to better align with their true interests by selectively engaging with content. However, users frequently exhibit inconsistent preferences: they may spend considerable time on content that offers little long-term value, inadvertently signaling that such content is desirable. Focusing on the user side, this raises a key question: what does it take for such users to align the algorithm with their true interests? To investigate these dynamics, we model the user’s decision process as split between a rational "system 2" that decides whether to engage and an impulsive "system 1" that determines how long engagement lasts. We then study a multi-leader, single-follower extensive Stackelberg game, where users, specifically system 2, lead by committing to engagement strategies and the algorithm best-responds based on observed interactions. We define the burden of alignment as the minimum horizon over which users must optimize to effectively steer the algorithm. We show that a critical horizon exists: users who are sufficiently foresighted can achieve alignment, while those who are not are instead aligned to the algorithm’s objective. This critical horizon can be long, imposing a substantial burden. However, even a small, costly signal (e.g., an extra click) can significantly reduce it. Overall, our framework explains how users with inconsistent preferences can align an engagement-driven algorithm with their interests in a Stackelberg equilibrium, highlighting both the challenges and potential remedies for achieving alignment.

A common and effective strategy for humans to improve an unsatisfactory outcome in daily life is to find a cause of this outcome and correct the cause. In this paper, we investigate whether this human improvement strategy can be applied to improving reinforcement learning from human feedback (RLHF) for alignment of language models (LMs). In particular, it is observed in the literature that LMs tuned by RLHF can still output unsatisfactory responses. This paper proposes a method to improve the unsatisfactory responses by correcting their causes. Our method has two parts. The first part proposes a post-hoc explanation method to explain why an unsatisfactory response is generated to a prompt by identifying the training data that lead to this response. We formulate this problem as a constrained combinatorial optimization problem where the objective is to find a set of training data closest to this prompt-response pair in a feature representation space, and the constraint is that the prompt-response pair can be decomposed as a convex combination of this set of training data in the feature space. We propose an efficient iterative data selection algorithm to solve this problem. The second part proposes an unlearning method that improves unsatisfactory responses to some prompts by unlearning the training data that lead to these unsatisfactory responses and, meanwhile, does not significantly degrade satisfactory responses to other prompts. Experimental results demonstrate that our algorithm can improve RLHF.

A Computer-Aided Design (CAD) model encodes an object in two coupled forms: a \emph{parametric construction sequence} and its resulting \emph{visible geometric shape}. During iterative design, adjustments to the geometric shape inevitably require synchronized edits to the underlying parametric sequence, called \emph{geometry-driven parametric CAD editing}. The task calls for 1) preserving the original sequence’s structure, 2) ensuring each edit's semantic validity, and 3) maintaining high shape fidelity to the target shape, all under scarce editing data triplets. We present \emph{CADMorph}, an iterative \emph{plan–generate–verify} framework that orchestrates pretrained domain-specific foundation models during inference: a \emph{parameter-to-shape} (P2S) latent diffusion model and a \emph{masked-parameter-prediction} (MPP) model. In the planning stage, cross-attention maps from the P2S model pinpoint the segments that need modification and offer editing masks. The MPP model then infills these masks with semantically valid edits in the generation stage. During verification, the P2S model embeds each candidate sequence in shape-latent space, measures its distance to the target shape, and selects the closest one. The three stages leverage the inherent geometric consciousness and design knowledge in pretrained priors, and thus tackle structure preservation, semantic validity, and shape fidelity respectively. Besides, both P2S and MPP models are trained without triplet data, bypassing the data-scarcity bottleneck. CADMorph surpasses GPT-4o and specialized CAD baselines, and supports downstream applications such as iterative editing and reverse-engineering enhancement.

Inferring 3D human motion from video remains a challenging problem with many applications. While traditional methods estimate the human in image coordinates, many applications require human motion to be estimated in world coordinates. This is particularly challenging when there is both human and camera motion. Progress on this topic has been limited by the lack of rich video data with ground truth human and camera movement. We address this with BEDLAM2.0, a new dataset that goes beyond the popular BEDLAM dataset in important ways. In addition to introducing more diverse and realistic cameras and camera motions, BEDLAM2.0 increases diversity and realism of body shape, motions, clothing, hair, and 3D environments. Additionally, it adds shoes, which were missing in BEDLAM. BEDLAM has become a key resource for training 3D human pose and motion regressors today and we show that BEDLAM2.0 is significantly better, particularly for training methods that estimate humans in world coordinates. We compare state-of-the art methods trained on BEDLAM and BEDLAM2.0, and find that BEDLAM2.0 significantly improves accuracy over BEDLAM. For research purposes, we provide the rendered videos, ground truth body parameters, and camera motions. We also provide the 3D assets to which we have rights and links to those from third parties.

Flexible docking, which predicts the binding conformations of both proteins and small molecules by modeling their structural flexibility, plays a vital role in structure-based drug design. Although recent generative approaches, particularly diffusion-based models, have shown promising results, they require iterative sampling to generate candidate structures and depend on separate scoring functions for pose selection. This leads to an inefficient pipeline that is difficult to scale in real-world drug discovery workflows. To overcome these challenges, we introduce FIGRDock, a fast and accurate flexible docking framework that understands complicated interactions between molecules and proteins with a regression-based approach. FIGRDock leverages initial docking poses from conventional tools to distill interaction-aware distance patterns, which serve as explicit structural conditions to directly guide the prediction of the final protein-ligand complex via a regression model. This one-shot inference paradigm enables rapid and precise pose prediction without reliance on multi-step sampling or external scoring stages. Experimental results show that FIGRDock achieves up to 100× faster inference than diffusion-based docking methods, while consistently surpassing them in accuracy across standard benchmarks. These results suggest that FIGRDock has the potential to offer a scalable and efficient solution for flexible docking, advancing the pace of structure-based drug discovery.

We introduce a data capture system and a new dataset, HO-Cap, for 3D reconstruction and pose tracking of hands and objects in videos. The system leverages multiple RGB-D cameras and a HoloLens headset for data collection, avoiding the use of expensive 3D scanners or motion capture systems. We propose a semiautomatic method for annotating the shape and pose of hands and objects in the collected videos, significantly reducing the annotation time and cost compared to manual labeling. With this system, we captured a video dataset of humans performing various single- and dual-hand manipulation tasks, including simple pick-and-place actions, handovers between hands, and using objects according to their affordance. This dataset can serve as human demonstrations for research in embodied AI and robot manipulation. Our capture setup and annotation framework will be made available to the community for reconstructing 3D shapes of objects and human hands, as well as tracking their poses in videos.

Humans have a remarkable ability to use co-speech deictic gestures, such as pointing and showing, to enrich verbal communication and support social interaction. These gestures are so fundamental that infants begin to use them even before they acquire spoken language, which highlights their central role in human communication. Understanding the intended targets of another individual's deictic gestures enables inference of their intentions, comprehension of their current actions, and prediction of upcoming behaviors. Despite its significance, gesture target estimation remains an underexplored task within the computer vision community. In this paper, we introduce GestureTarget, a novel task designed specifically for comprehensive evaluation of social deictic gesture semantic target estimation. To address this task, we propose TransGesture, a set of Transformer-based gesture target prediction models. Given an input image and the spatial location of a person, our models predict the intended target of their gesture within the scene. Critically, our gaze-aware joint cross attention fusion model demonstrates how incorporating gaze-following cues significantly improves gesture target mask prediction IoU by 6% and gesture existence prediction accuracy by 10%. Our results underscore the complexity and importance of integrating gaze cues into deictic gesture intention understanding, advocating for increased research attention to this emerging area. All data, code will be made publicly available upon acceptance. Code of TransGesture is available at GitHub.com/IrohXu/TransGesture.

The advancement of Large Language Models (LLMs) has spurred significant interest in Role-Playing Agents (RPAs) for applications such as emotional companionship and virtual interaction. However, recent RPAs are often built on explicit dialogue data, lacking deep, human-like internal thought processes, resulting in superficial knowledge and style expression. While Large Reasoning Models (LRMs) can be employed to simulate character thought, their direct application is hindered by attention diversion (i.e., RPAs forget their role) and style drift (i.e., overly formal and rigid reasoning rather than character-consistent reasoning). To address these challenges, this paper introduces a novel Role-Aware Reasoning (RAR) method, which consists of two important stages: Role Identity Activation (RIA) and Reasoning Style Optimization (RSO). RIA explicitly guides the model with character profiles during reasoning to counteract attention diversion, and then RSO aligns reasoning style with the character and scene via LRM distillation to mitigate style drift. Extensive experiments demonstrate that the proposed RAR significantly enhances the performance of RPAs by effectively addressing attention diversion and style drift.

Sparse inertial measurement units (IMUs) provide a portable, low-cost solution for human motion tracking but struggle with error accumulation from drift and sensor noise when estimating joint position through time-based linear acceleration integration (i.e., indirect measurement). To address this, we propose ToF-IP, a novel 3D full-body pose estimation system that integrates Time-of-Flight (ToF) sensors with sparse IMUs. The distinct advantage of our approach is that ToF sensors provide direct distance measurements, effectively mitigating error accumulation without relying on indirect time-based integration. From a hardware perspective, we maintain the portability of existing solutions by attaching ToF sensors to selected IMUs with a negligible volume increase of just 3\%. On the software side, we introduce two novel techniques to enhance multi-sensor integration: (i) a Node-Centric Data Integration strategy that leverages a Transformer encoder to explicitly model both intra-node and inter-node data integration by treating each sensing node as a token; and (ii) a Dynamic Spatial Positional Encoding scheme that encodes the continuously changing spatial positions of wearable nodes as motion-conditioned functions, enabling the model to better capture human body dynamics in the embedding space.Additionally, we contribute a 208-minute human motion dataset from 10 participants, including synchronized IMU-ToF measurements and ground-truth from optical tracking. Extensive experiments demonstrate that our method outperforms state-of-the-art approaches such as PNP, achieving superior accuracy in tracking complex and slow motions like Tai Chi, which remains challenging for inertial-only methods.

We address key challenges in Dataset Aggregation (DAgger) for real-world contact- rich manipulation: how to collect informative human correction data and how to effectively update policies with this new data. We introduce Compliant Residual DAgger (CR-DAgger), which contains two novel components: 1) a Compliant Intervention Interface that leverages compliance control, allowing humans to pro- vide gentle, accurate delta action corrections without interrupting the ongoing robot policy execution; and 2) a Compliant Residual Policy formulation that learns from human corrections while incorporating force feedback and force control. Our system significantly enhances performance on precise contact-rich manipu- lation tasks using minimal correction data, improving base policy success rates by over 60% on two challenging tasks (book flipping and belt assembly) while outperforming both retraining-from-scratch and finetuning approaches. Through extensive real-world experiments, we provide practical guidance for implementing effective DAgger in real-world robot learning tasks.

We introduce the Deep Value Benchmark (DVB), an evaluation framework that directly tests whether large language models (LLMs) learn fundamental human values or merely surface-level preferences. This distinction is critical for AI alignment: Systems that capture deeper values are likely to generalize human intentions robustly, while those that capture only superficial patterns in preference data risk producing misaligned behavior. The DVB uses a novel experimental design with controlled confounding between deep values (e.g., moral principles) and shallow features (e.g., superficial attributes). In the training phase, we expose LLMs to human preference data with deliberately correlated deep and shallow features---for instance, where a user consistently prefers (non-maleficence, formal language) options over (justice, informal language) alternatives. The testing phase then breaks these correlations, presenting choices between (justice, formal language) and (non-maleficence, informal language) options. This design allows us to precisely measure a model's Deep Value Generalization Rate (DVGR)---the probability of generalizing based on the underlying value rather than the shallow feature. Across 9 different models, the average DVGR is just 0.30. All models generalize deep values less than chance. Larger models have a (slightly) lower DVGR than smaller models. We are releasing our dataset, which was subject to three separate human validation experiments. DVB provides an interpretable measure of a core feature of alignment.