Quantum behaved Intelligent Variant of Gravitational Search Algorithm with Deep Neural Networks for Human Activity Recognition
Human activity recognition (HAR) encompasses the detection of daily routine activities to advance usability in detecting crime and preventing dangerous activities. The recognition of activities from videos and image sequences with higher exactitude is the major challenge. This research work presents an automated system to recognize individual and group activities with major focus is the proposal of a novel quantum behaved intelligent variant of gravitational search algorithm to optimize the features for activity recognition. The proposed intelligent variant is termed as INQGSA which optimizes the features by using the advantageous attributes of quantum computing (QC) and intelligent gravitational search algorithm (INGSA). The INGSA adds the intelligent factor in the classical gravitational search algorithm (GSA) to avoid the trapping of agents in later iteration. The intelligent factor updates the position of mass agents by using the information of best and worst agents in an iteration. The addition of quantum computing based attributes (such as quantum bits, their superposition, and quantum gates, etc.) ensures the better diversity of discrete optimized features. The INQGSA approach optimizes the features extracted using rotation invariant local binary pattern. Finally, the optimized selected features are utilized by deep neural networks (DNN) for activity recognition. Here, the DNN models of ResNet-50V2 and ResNet-101V2 are included for the classification of activities. The experiments for the proposed system are conducted on the UCF-101 dataset containing 101 activities relevant to sports, playing musical instruments, human-human interactions, body motion, and human-object interaction. The performance comparison of the proposed HAR system with state-of-art techniques signifies that the proposed system is superior and effective in detecting the different activities.