Multi-Camera Self-Calibration in Sports Motion Capture: Leveraging Human and Stick Poses

2026-04-19

AI Summary

This work addresses the high cost of multi-camera systems in sports motion capture, which typically rely on specialized tools for extrinsic calibration. We propose a novel self-calibration method that eliminates the need for calibration objects by jointly leveraging scale-ambiguous human keypoints and rigid rods of known length—such as golf or hockey sticks—as complementary geometric constraints from synchronized videos. Through a three-stage optimization pipeline, our approach simultaneously recovers camera extrinsics, 3D human poses, and rod trajectories, resolving global scale ambiguity using the known rod length. We introduce the first multi-camera self-calibration dataset tailored for rod-based sports and demonstrate significant reductions in rotation and translation errors across configurations with 3 to 10 cameras, achieving accurate and robust markerless extrinsic estimation.

Technology Category

• Computer Vision: Motion & Tracking
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Multi-camera systems are widely employed in sports to capture the 3D motion of athletes and equipment, yet calibrating their extrinsic parameters remains costly and labor-intensive. We introduce an efficient, tool-free method for multi-camera extrinsic calibration tailored to sports involving stick-like implements (e.g., golf clubs, bats, hockey sticks). Our approach jointly exploits two complementary cues from synchronized multi-camera videos: (i) human body keypoints with unknown metric scale and (ii) a rigid stick-like implement of known length. We formulate a three-stage optimization pipeline that refines camera extrinsics, reconstructs human and stick trajectories, and resolves global scale via the stick-length constraint. Our method achieves accurate extrinsic calibration without dedicated calibration tools. To benchmark this task, we present the first dataset for multi-camera self-calibration in stick-based sports, consisting of synthetic sequences across four sports categories with 3 to 10 cameras. Comprehensive experiments demonstrate that our method delivers SOTA performance, achieving low rotation and translation errors. Our project page: https://fandulu.github.io/sport_stick_multi_cam_calib/.

Problem

• multi-camera calibration
• sports motion capture
• extrinsic parameters
• self-calibration
• stick-like implements

Innovation

• multi-camera self-calibration
• sports motion capture
• stick pose
• human keypoints
• scale disambiguation