Understanding human activitythrough computer vision.
STELLA observes human movement, hand interaction, and objects during space experiments to understand what is happening — and whether the procedure is being followed.
Space experiments
are sequences.
Astronauts do not simply move through an experiment. They perform a sequence of deliberate actions — reaching, grasping, moving, placing, and withdrawing.
STELLA uses computer vision to interpret those actions in context, turning visual observations into meaningful experimental events.
See STELLA working ↗See what
STELLA sees.
Two ways to experience the system — interact with the live perception pipeline or explore recorded experiments.
From perception
to understanding.
STELLA combines local computer vision, human activity recognition, interaction detection, and procedure validation into a single offline experiment runtime.
Live visual input from the experiment environment.
Pose, hand, and object perception running locally.
Movement and interaction evidence are interpreted as structured activity.
Detected actions are evaluated against the active experiment procedure.
Everything
behind STELLA.
Explore the implementation, research material, demonstrations, data, and project documentation.
Multiple signals.
One understanding.
STELLA combines several visual perception layers to understand activity instead of treating every movement as an isolated classification.
INTELLIGENT HUMAN ACTIVITY RECOGNITION