Abstract: This paper presents GPU based CUDA framework that is efficiently and accurately detect unattended object from surveillance video. The system focuses on the problem of finding the unattended object in public places such as shopping mall, airport, railway station etc. In a recent year, GPU has attracted the attention of many application developers as powerful massively parallel system. CUDA as a general purpose parallel computing architectures that makes GPU is an appealing choice to solve many complex computational problems in a more effective way. The processing of surveillance video is computationally intensive. This paper describes parallel implementation of video object detection algorithms like Gaussian Mixture Model(GMM) for background modeling, morphological operation for post processing. SVM classifier is used for unattended object detection. Experimental evaluation shows that parallel GPU implementation achieves significant speedup for GMM and Morphological operations when compared to sequential implementation running on Intel Processor.

Keywords: GPU, CUDA, Parallel computing, unattended object.

PDF | DOI: 10.17148/IJARCCE.2018.765

Open chat
Chat with IJARCCE