Computer vision approaches based on deep learning and neural networks: Deep neural networks for video analysis of human pose estimation

Deep architectures with convolution structure have been found highly effective and commonly used in computer vision. With the introduction of Graphics Processing Unit (GPU) for general purpose issues, there has been an increasing attention towards exploiting GPU processing power for deep learning algorithms. Also, large amount of data online has made possible to train deep neural networks efficiently. The aim of this paper is to perform a systematic mapping study, in order to investigate existing research about implementations of computer vision approaches based on deep learning algorithms and Convolutional Neural Networks (CNN). We selected a total of 119 papers, which were classified according to field of interest, network type, learning paradigm, research and contribution type. Our study demonstrates that this field is a promising area for research. We choose human pose estimation in video frames as a possible computer vision task to explore in our research. After careful studying we propose three different research direction related to: improving existing CNN implementations, using Recurrent Neural Networks (RNNs) for human pose estimation and finally relying on unsupervised learning paradigm to train NNs.