PoseForge: Editable Pose Analytics for AI-Assisted Sports Coaching

Athletic coaching often relies on subjective observation or expensive, laboratory-bound motion capture systems. While video analysis is common, it lacks the ability to quantify biomechanical movement or simulate corrective adjustments. To address this, the authors developed PoseForge, a visual analytics tool designed for grassroots and academy-level coaching. The system uses a markerless pose estimation pipeline to reconstruct 3D skeletal motion from monocular video. It then computes interpretable kinematic metrics—such as feet gap and joint angles—and compares them against scientifically derived norms.

A key innovation of PoseForge is its support for "what-if" analysis. Coaches can modify a player's reconstructed pose using mouse-based manipulation or natural-language instructions. To ensure these corrections are physically realistic, the system employs an inverse-kinematics (IK) engine that enforces anatomical joint limits. This allows coaches to demonstrate the impact of specific adjustments (e.g., "increase feet gap by 10 cm") in real-time, providing a bridge between abstract biomechanical data and actionable, visual feedback.

The researchers conducted a formative interview study with eleven cricket experts to identify coaching pain points, followed by a summative evaluation with the same group. Participants found the system effective for diagnosing movement issues and praised the ability to simulate corrections, noting that the tool could significantly improve the clarity and credibility of coaching feedback. The study highlights that browser-based, AI-assisted tools can effectively democratize high-fidelity motion analysis, making professional-grade biomechanical insights accessible in resource-constrained environments.