Comparative Analysis and Implementation of Different Human Detection Techniques

Object Detection is used in making several smart surveillance applications which are used in detecting and tracking suspicious activities. Object Detection also plays a very crucial role in several latest inventions like the Google self-driving car. Over the past decade several image processing algorithms have evolved which have been used for human detection after preprocessing of the images. But in several cases these algorithms were found to be less accurate and more time taking. With the advent of the era of Machine Learning, the focus has gradually shifted towards computer vision and deep learning methodologies for human detection. Deep learning algorithms are far more accurate than the traditional methodologies as they employ feature extraction from the images followed by classification according to the dataset provided to them, thus enabling more accurate detection than their ancestors. Nevertheless its also true that some deep learning algorithms like the different versions of R-CNN and YOLO take much more processing time than the traditional methodologies. In our paper we have compared the performance of some traditional and some deep learning algorithms in different scenarios.