# Crowd4D: Scene-Aware Monocular 4D Crowd Reconstruction

arXiv: https://arxiv.org/abs/2607.19517 · HTML: https://mlsanity.com/paper/2607.19517/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 28 of its 30 references that are in the map.
- First posted on arXiv 21 Jul 2026.

## Builds on (most cited first, 28 of 28)

- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [SAM 2: Segment Anything in Images and Videos](https://mlsanity.com/paper/2408.00714/) (2024)
- [Expressive Body Capture: 3D Hands, Face, and Body from a Single Image](https://mlsanity.com/paper/1904.05866/) (2019)
- [VGGT: Visual Geometry Grounded Transformer](https://mlsanity.com/paper/2503.11651/) (2025)
- [YOLOX: Exceeding YOLO Series in 2021](https://mlsanity.com/paper/2107.08430/) (2021)
- [Humans in 4D: Reconstructing and Tracking Humans with Transformers](https://mlsanity.com/paper/2305.20091/) (2023)
- [π3: Permutation-Equivariant Visual Geometry Learning](https://mlsanity.com/paper/2507.13347/) (2025)
- [Effective Whole-body Pose Estimation with Two-stages Distillation](https://mlsanity.com/paper/2307.15880/) (2023)
- [Monocular, One-stage, Regression of Multiple 3D People](https://mlsanity.com/paper/2008.12272/) (2020)
- [Decoupling Human and Camera Motion from Videos in the Wild](https://mlsanity.com/paper/2302.12827/) (2023)
- [TRAM: Global Trajectory and Motion of 3D Humans from in-the-wild Videos](https://mlsanity.com/paper/2403.17346/) (2024)
- [Multi-HMR: Multi-Person Whole-Body Human Mesh Recovery in a Single Shot](https://mlsanity.com/paper/2402.14654/) (2024)
- [PromptHMR: Promptable Human Mesh Recovery](https://mlsanity.com/paper/2504.06397/) (2025)
- [Visual Imitation Enables Contextual Humanoid Control](https://mlsanity.com/paper/2505.03729/) (2025)
- [CameraHMR: Aligning People with Perspective](https://mlsanity.com/paper/2411.08128/) (2024)
- [SMAP: Single-Shot Multi-Person Absolute 3D Pose Estimation](https://mlsanity.com/paper/2008.11469/) (2020)
- [PANDA: A Gigapixel-level Human-centric Video Dataset](https://mlsanity.com/paper/2003.04852/) (2020)
- [AiOS: All-in-One-Stage Expressive Human Pose and Shape Estimation](https://mlsanity.com/paper/2403.17934/) (2024)
- [Human3R: Everyone Everywhere All at Once](https://mlsanity.com/paper/2510.06219/) (2025)
- [Reconstructing People, Places, and Cameras](https://mlsanity.com/paper/2412.17806/) (2024)
- [Joint Optimization for 4D Human-Scene Reconstruction in the Wild](https://mlsanity.com/paper/2501.02158/) (2025)
- [CIMI4D: A Large Multimodal Climbing Motion Dataset under Human-scene Interactions](https://mlsanity.com/paper/2303.17948/) (2023)
- [Reconstructing Groups of People with Hypergraph Relational Reasoning](https://mlsanity.com/paper/2308.15844/) (2023)
- [Synergistic Global-space Camera and Human Reconstruction from Videos](https://mlsanity.com/paper/2405.14855/) (2024)
- [BoostTrack++: using tracklet information to detect more objects in multiple object tracking](https://mlsanity.com/paper/2408.13003/) (2024)
- [Crowd3D: Towards Hundreds of People Reconstruction from a Single Image](https://mlsanity.com/paper/2301.09376/) (2023)
- [DyCrowd: Towards Dynamic Crowd Reconstruction from a Large-scene Video](https://mlsanity.com/paper/2508.12644/) (2025)
- [RCR: Robust Crowd Reconstruction with Upright Space from a Single Large-scene Image](https://mlsanity.com/paper/2411.06232/) (2024)

## Abstract (arXiv)

Recovering scene-consistent 4D crowd motion from monocular video in large-scale scenes remains challenging due to severe depth ambiguity and complex scene geometry. Existing monocular crowd reconstruction methods typically rely on single-plane assumptions, leading to unreliable metric scale and spatial drift under complex terrain. We propose Crowd4D, the first scene-aware 4D crowd reconstruction framework that jointly optimizes the crowd and scene from a monocular RGB video in large-scale scenes. Crowd4D explicitly incorporates scene geometry and ensures consistency across image and scene spaces via a multi-stage optimization strategy. A key bottleneck of this task lies in accurate human-scene alignment, particularly in scale and position. However, human and scene reconstructions are typically decoupled. To address this, we introduce the Human-Scene Interaction Proxy, abbreviated as HSIP, as an intermediate representation derived from Scene Interaction Point Clouds and a Scene Interaction Surface, abbreviated as SIPC and SIS. These representations encode explicit scene-aware geometric priors and redefine the optimization space for large-scale monocular 4D crowd reconstruction. To further improve temporal stability under occlusions, we introduce Crowd Structural Coherence Regularization, abbreviated as CSCR, which leverages HSIP-based spatial priors to impose soft temporal consistency on pairwise relative displacements and directions within local crowd neighborhoods. Extensive experiments demonstrate that Crowd4D consistently outperforms existing state-of-the-art methods and enables robust monocular 4D crowd reconstruction in complex, large-scale real-world scenes.