Comparative Analysis of Hand-Coded and Automated Feature Based Algorithms for Human Detection in Embedded Application

• Published in: 2018 3rd International Conference for Convergence in Technology (I2CT) pp. 1 - 6
• Main Authors: Agarwal, Naman, Suthar, Sanjay, Jain, Harshita, Godara, Pratiti, Vaidya, Bhaumik
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.04.2018
• Subjects: Classification algorithms; Deep Learning; Detectors; Feature extraction; Haar Cascade; Histogram of Oriented Gradient; Histograms; Machine Learning; Microsoft Windows; Object detection; Single Shot Multibox Detector; Support Vector Machine; Support vector machines
• DOI: 10.1109/I2CT.2018.8529577

Human detection in the video frame is a difficult task and time consuming process. Inherent characteristics like posture, mien, physique and alien features like radiance, blockage, cluster, litter are prominent factors effecting the task. The main aim of this paper is to implement, compute and compare different algorithms in unfavorable conditions and conclude the output parameters of the algorithm on real time videos. In this paper, three feature based algorithms Single Shot Multibox Detector (SSD), Haar Cascade and Histogram of Oriented Gradient (HOG) are computed. The comparison of this algorithms in real life scenarios and on different platforms are being computed to find out which algorithm is suitable for an embedded application.