cc-collection

High-performance sensor fusion library for motion capture data collection. Combines Sony Mocopi IMU sensors with MediaPipe pose estimation using Extended Kalman Filtering for ML training data generation. - **Multi-Sensor Fusion**: Combines Mocopi (6 IMU sensors) with MediaPipe (camera-based) tracking - **Extended Kalman Filter**: 13D state estimation per limb with adaptive noise modeling - **Real-time Processing**: 60Hz+ fusion with sub-millisecond latency - **Beat Synchronization**: Phase-aligned motion vectors for music-driven applications - **Motion Transforms**: 25D and 63D feature vectors for ML training - **Cross-Platform**: Rust core with Python bindings (PyO3) and WebAssembly support | Type | Description | |------|-------------| | `LimbId` | Enum for 14 tracked limbs (Hip, Head, LeftShoulder, etc.) | | `LimbState` | Per-limb state: position, quaternion, velocity, confidence | | `FusedSkeleton` | Complete skeleton with all limb states | | `FusionMode` | Current mode: None, MocopiOnly, MediaPipeOnly, Fused | | `DataSource` | Source tracking: Mocopi, MediaPipe, Fused, Interpolated | | Component | Description | |-----------|-------------| | `FusionConfig` | Configuration: session_id, min_confidence, smoothing | | `FusionEngine` | Main processor with EKF-based fusion | | `FusionStats` | Runtime statistics: frame count, sync status | | Transform | Output | Description | |-----------|--------|-------------| | `To25DTransform` | 25D vector | Core motion features for ML training | | `To63DTransform` | 63D vector | Extended features with face/hands |