Releases: wearable-motion-capture/arm-pose-estimation
Releases · wearable-motion-capture/arm-pose-estimation
Kalman Cleanup And Activity Rec Prep
- Documentation and transformation cleanup for Kalman classes
- Introduce One-Hot data handling for activity recognition
- Rework estimator structure to "run-thread-calls" for clearer inheritance and more intuitive example scripts
Phone Mode Updates
- Updated Kalman model for Pocket mode
- Enabled stochastic forward passes for the Upper Arm mode. The old one is moved to experimental scripts as "raw_upper_arm".
- Fixed pip-install bug where the kalman model is not copied to the target directory
Phone Pocket handling
- Added a Kalman Filter approach to allow free body orientation changes tracked by the phone
- See associated Workshop Paper for this release
- New message structure and new models for old modes
first basic release
- The first basic release provides streaming and record examples for the modes WatchOnly and PhoneUpperArm.
- Additional functionalities remain in the experimental scripts category until they are thoroughly documented and tested