A Real-Time Visual Tracking Approach Using Sparse Autoencoder and Extreme Learning Machine

作者：Dai, Liang; Zhu, Yuesheng; Luo, Guibo; He, Chao; Lin, Hanchi

来源：Unmanned Systems, 2015, 3(4): 267-275.

DOI：10.1142/S2301385015400038

• Auto encoders
• Computational burden
• Computational costs
• Extreme learning machine
• Real time
• Single-hidden-layer neural networks
• Visual Tracking
• Visual tracking algorithm

Visual tracking algorithm based on deep learning is one of the state-of-the-art tracking approaches. However, its computational cost is high. To reduce the computational burden, in this paper, A real-time tracking approach is proposed by using three modules: a single hidden layer neural network based on sparse autoencoder, a feature selection for simplifying the network and an online process based on extreme learning machine. Our experimental results have demonstrated that the proposed algorithm has good performance of robust and real-time.

• 出版日期2015-10-1
• 单位北京大学