Human detections using Beagle board-XM

Abstract

In this paper, we describe implementation of human detection system that detects the presence of humans in the static images on DM3730 processor to optimize the algorithms for high performance. The purpose of such a model is to monitor surveillance cameras continuously where it is difficult for human operators. Human bodies are non-rigid and highly articulated hence detecting human bodies based on appearance is more difficult than detecting other objects. Human detector usually includes learning phase and detection phase. In the learning phase, Support Vector Machine (SVM) learning will be done using Histogram of Oriented Gradients (HOG) feature vectors of training data set. Training data set consists of positive (human) and negative (non-human) images. In the detection phase, Human/non-human classification will be done on the test image based on SVM classifier. The algorithms are benchmarked on the BeagleBoard xM based on low-power Texas Instruments (TI) DM3730 ARM Cortex-A8 processor. Functions and library in OpenCV which developed by Intel Corporation was utilized for building the human tracking algorithms.