Skip to main content

Process Markers

Optical Marker based Body Tracking​

This project provides a set of core functions to solve for body motions based on optical marker inputs. It supports all PC OSes (untested on mobile). The marker_tracker lib contains core functionalities for downstream applications to build on. Demo applications are provided to show how they can be used to build your data processing pipeline. process_markers_app solves for body motion given an input marker sequence, with or without an existing calibrated skeleton. refine_motion runs smoothing as a post process to fill in missing data from input. They can be used to batch process mocap data in a python script.

info

The Momentum ecosystem implicitly operates in centimeter. If you are working with c3d files, we will do the unit conversion based on the stored unit on file. However, if you are using our API with your own data, make sure to convert them into cm. We also assume a Y-up coordinate system, which is not the industry convention (i.e., Z-up).

Example use cases​

Get the full list of options for each application with -h or --help argument. 02_01.c3d is an example input file used by the default config files. Note that a config file can be used together with command line options. The command line overwrites values in the config file.

Track a marker sequence without a calibrated model.​

The first step in tracking a marker file is to calibrate the subject's proportions and the markers' placement. It requires a .locators file that defines a template of marker layout on the body. We have a template file with common layouts from Vicon and OptiTrack. There is usually a Range-of-Motion (ROM) sequence captured for this calibration purpose.

Use a config file:

pixi run process_markers -c process_markers_calib.config

Setting the calibrate option to true will first calibrate the skeleton and the marker layout, then use the calibrated model for motion tracking.

Track a marker sequence with a calibrated model.​

The tracking result from the above calibration step contains the calibrated model, and it can then be used to track other motion data from the same subject, without running the calibration step again. We currently only support saving/loading calibrated models in .glb format.

Use a config file:

pixi run process_markers -c process_markers_tracking.config

Use cli arguments:

pixi run process_markers -i input.c3d -o tracked.glb --model calibrated_model.glb --calibrate false