Complex Behavior Recognition Based on Convolutional Neural Network: A Survey

Abstract

Behavior recognition is an important research direction in computer vision. The behavior recognition based on convolutional neural network has become a research hotspot in recent years. The methods based on convolutional neural network can extract features directly from video data, reduce the difference of temporal domain and the influence of spatial complexity. At present, the simple behavior recognition based on convolutional neural network has been solved basically. However, the complex behavior recognition based on convolutional neural network still faces many difficulties. In this paper, the issues of spatial dependencies and time dependencies in complex behavior recognition are discussed. Then convolutional neural network applying to complex behavior recognition is analyzed in detail from time, space, and spatio-temporal aspects following research progress. Finally, the future development of complex behavior recognition based on convolutional neural network is indicated.