.png){kind=icon}
.png){kind=icon}
Return
Visual gait recognition based on convolutional block attention network
.png){kind=icon}
PRE
AI
DOI：10.1007/s11042-022-12831-1.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Gait recognition has many advantages, such as non-invasive and easy to recognize from a long distance. It has a broad application prospect in the field of human identification. However, due to the sensitivity of gait recognition to sample collection perspective, there is no effective solution to cross-view gait recognition. To solve the above problems, this paper proposes a visual gait recognition method based on convolutional block attention network. Firstly, the method takes frame by frame gait energy images as the input, and makes each frame go through the attention neural network with the same structure to extract whole gait features. Then, the whole gait features are divided into two parts to train the network model and recognize the unknown gaits. Finally, the experimental results on the open-accessed gait data sets CASIA-B and OU-ISIR-MVLP show that the proposed method is more accurate than the existing methods, i.e., the Rank 1 rate is increased by at least 2.8% on CASIA gait dataset B, and more than 10% on OU-ISIR MVLP dataset.
Keywords:
Gait recognition
Deep learning
Convolutional block attention network
Human identification
Journal
IF:
3
Papers:
1.9W
Citations:
3.2W