Violent crowd behavior detection using deep learning and compressive sensing
Mingliang Gao, Jun Jiang, Lixiu Ma, Shuwen Zhou, Guofeng Zou, Jinfeng Pan, Zheng Liu · 2019
In this paper, we propose a new crowd analysis method fusing deep learning network and compressive sensing for violent behavior detection. To this aim, a novel hybrid random matrix (HRM) is constructed and is proved to satisfy the restricted isometry property. The high-dimensional features can be projected to a low-dimensional space via the HRM. Furthermore, a deep neural network is developed for extracting crowd behaviour representations based on reduced dimension features. Finally, the learnt deep features are used for classification. Experimental results demonstrate that the proposed method is effective and efficient in violent behavior detection, and is on par with or better than the state-of-the-art methods.