Year: 2013
Type: dissertation
Abstract: Modeling and recognizing human behaviors in a visual surveillance task is receiving\nincreasing attention from computer vision and machine learning researchers. Such a system\nshould deal in particularly with detecting when interactions between people occur and\nclassifying the type of interaction.\nIn this work we study a flexible model for detecting human interactions. This has\nbeen done by detecting the people in the scene and retrieving their corresponding pose and\nposition sequentially in each frame of the video. To achieve this goal our work relies on\nrobust object detection algorithm which is based on discriminatively trained part based\nmodels to detect the human bodies in videos. We apply a ‘Gaussian Mixture Models based’\nmethod for background subtraction and human segmentation. The output from the\nsegmentation method which is labeled human body is combined with the background\nsubtraction methods to obtain a bounding box around each person in images to improve the\ntask of human body pose detection.\nTo gain more precise pose detection models, we trained the algorithm on large,\nchallenging but reliable dataset (PASCAL 2010). Our method is applied in home-made\ndatabase comprising depth data from Kinect sensors. After successfully getting in every\nimage sequence the corresponding label for each person as well as their pose and position,\nunderstanding of human motion comes naturally which is an important step towards human\ninteraction analysis.
Author: Maedeh Aghaei Gavari
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W96094990