期刊论文

Bing Yang

英文

Neurocomputing

0925-2312

2021

436

210-220

10.1016/j.neucom.2020.12.090

To measure the similarities of adjacent head poses, we introduce the cosine function to compute the HPS of two head image features. The most representative features of one head image are extracted using the last fully-connected layer of a pretrained network. In Fig. 2(a), the central image is the (0°, 0°) angle pose image I0. With the increase of 15° in the pitch direction, one of the (0°, 15°) angle pose in the pitch direction is shown in image I1 as in Fig. 2(b). For the two IR images I0 and I

本校为其他机构

Head pose estimation (HPE) under active infrared (IR) illumination has attracted much attention in the fields of computer vision and machine learning. However, IRHPE often suffers from the problems of low-quality IR images and ambiguous head pose. To tackle these issues, we propose a novel nonuniform Gaussian-label distribution learning network (NGDNet) for the HPE task. First, we reveal the essential properties from two different perspectives: 1) two head pose images change differently in pitch and yaw directions with the same angle increasing on the central pose; 2) the IR head pose variatio...