We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Scale estimation-based visual tracking with optimized convolutional activation features.
- Authors
Guo, Qiang; Cao, Xuefei; Zou, Qinglong
- Abstract
Convolutional neural networks (CNNs) have proven very effective for learning features in visual tracking. While working effectively, it is still very challenging due to the scale variations and deformation, which may cause inconsecutive tracking trajectory and distraction. In this paper, pre-train deep learning network architecture is adopted for visual tracking, by introducing a spectral pooling in the network. Then, we propose an algorithm which, by interpreting scale correlation filters as the corresponding function of convolution filters in deep neural networks, exploits multilevel CNNs activation features into a new tracking framework. Finally, two-stage fine-tuning is then introduced for updating the model to keep long-time tracking. We test the proposed tracking method on large-scale benchmark sequences. Experimental results illustrate the effectiveness of the proposed algorithm compared with other state-of-the-art methods.
- Subjects
VISUAL learning; OBJECT tracking (Computer vision); DEEP learning
- Publication
Machine Vision & Applications, 2019, Vol 30, Issue 7/8, p1263
- ISSN
0932-8092
- Publication type
Article
- DOI
10.1007/s00138-019-01049-1