- Students Info
(Link to project documentation folder to be placed here)
Falls are the leading cause of injury and death among older adults .
Many elderlies are living alone and in a case of falling has no ability to call for help.
Existing solutions such as wearable sensors and floor sensors are expensive and uncomfortable.
Using depth data is less invasive than using RGB data and active under all lighting conditions.
Previous works- Approaches to Fall Detection
Fall detection based on Skeleton recognition
– Use joints velocity and distance from the floor in order to determine if a fall occurred.
Fall detection based on depth segmentation.
– use segmentation to create a bounding box around the user and detect the fall based on the bounding box dimensions and velocity.
Previous works- Example of skeleton base fall detection
Extract joints and floor plain.
Calculate the joints velocity and distance from the floor.
Fall is detected when the distance and the velocity are crossing a certain threshold values.
Previous works- Example of depth segmentation base fall detection
A bounding box is detected based on segmentation and motion.
Calculating the center of mass velocity.
A fall is detected based on the box dimensions and velocity.