Anymate: A Dataset and Baselines for Learning 3D Object Rigging

2025-05-09

AI Summary

Automatic rigging of complex-geometry 3D models suffers from poor generalizability of traditional geometry-driven heuristics and limited scalability of data-driven approaches due to scarcity of expert annotations. Method: We introduce Anymate—the first large-scale, expert-annotated rigging dataset (230K assets), 70× larger than prior benchmarks—and propose a three-stage end-to-end learning framework that sequentially predicts joint locations, skeletal topology, and skinning weights; the modular design enables full automation and granular evaluation. Our architecture integrates graph neural networks, point-cloud encoders, and weighted regression modules for robust, sequential prediction. Contribution/Results: Our method achieves significant improvements over state-of-the-art on Anymate, establishing a new benchmark. Both code and dataset are publicly released to foster standardized, reproducible research in automatic rigging.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Rigging and skinning are essential steps to create realistic 3D animations, often requiring significant expertise and manual effort. Traditional attempts at automating these processes rely heavily on geometric heuristics and often struggle with objects of complex geometry. Recent data-driven approaches show potential for better generality, but are often constrained by limited training data. We present the Anymate Dataset, a large-scale dataset of 230K 3D assets paired with expert-crafted rigging and skinning information -- 70 times larger than existing datasets. Using this dataset, we propose a learning-based auto-rigging framework with three sequential modules for joint, connectivity, and skinning weight prediction. We systematically design and experiment with various architectures as baselines for each module and conduct comprehensive evaluations on our dataset to compare their performance. Our models significantly outperform existing methods, providing a foundation for comparing future methods in automated rigging and skinning. Code and dataset can be found at https://anymate3d.github.io/.

Problem

• Automating 3D rigging and skinning for complex geometries
• Addressing limited training data for data-driven rigging methods
• Improving accuracy of joint, connectivity, and skinning predictions

Innovation

• Large-scale dataset with 230K 3D assets
• Learning-based auto-rigging framework with three modules
• Outperforms existing methods in automated rigging