filmov
tv
Gaussian Mixture Model based Object Detection and Tracking using Dynamic Patch Estimation
Показать описание
This video shows the implementation of Gaussian
Mixture Model (GMM) with dynamic patch estimation for real-
time detection and tracking of a known object. We have devised
a novel architecture that detects the object of interest, estimates
its 3-D position with respect to the quad-rotor using Extended
Kalman Filter (EKF) and finally generates the control output
to the quad-rotor to keep a predefined distance from the target.
The proposed object detection algorithm is capable of tracking
the object with high Frame Per Second (FPS) for closer objects
as well.
Mixture Model (GMM) with dynamic patch estimation for real-
time detection and tracking of a known object. We have devised
a novel architecture that detects the object of interest, estimates
its 3-D position with respect to the quad-rotor using Extended
Kalman Filter (EKF) and finally generates the control output
to the quad-rotor to keep a predefined distance from the target.
The proposed object detection algorithm is capable of tracking
the object with high Frame Per Second (FPS) for closer objects
as well.