HOG Based Contriving Individual detection

Journal: GRENZE International Journal of Engineering and Technology
Authors: Sowmya K S
Volume: 3 Issue: 3
Grenze ID: 01.GIJET.3.3.149 Pages: 191-195


The workplace of building construction is a cluttered enviornment which makes the automated detection of workers challenging. Challenges include placement of cameras, variation of lighting conditions, occlusion etc. In this paper HOG is used as the feature detector and Linear SVM is used to classify the person from the background.The algorithm is tested using INRIA database and also on real time videos captured from construction sites.The detection rate is 99% for the INRIA database and 90% for the real time vieos captured.