UCL Discovery Stage
UCL home » Library Services » Electronic resources » UCL Discovery Stage

Hourglass Face Detector for Hard Face

Yu, Z; Yin, J; Zhang, Q; Yang, W; Xue, J-H; Liao, Q; (2021) Hourglass Face Detector for Hard Face. In: 2021 International Joint Conference on Neural Networks (IJCNN). IEEE Green open access

[thumbnail of ZijunYu-IJCNN2021-accepted.pdf]
Preview
Text
ZijunYu-IJCNN2021-accepted.pdf - Accepted Version

Download (3MB) | Preview

Abstract

Face detection is an upstream task of facial image analysis. In many real-world scenarios, we need to detect small, occluded or dense faces that are hard to detect, but hard face detection is a challenging task in particular considering the balance between accuracy and inference speed for real-world applications. This paper proposes an Hourglass Face Detector (HFD) for hard face by developing a deep one-stage fully-convolutional hourglass network, which achieves an excellent balance between accuracy and inference speed. To this end, the HFD firstly shrinks a feature map by a series of stridden convolutional layers rather than pooling layers, so that useful subtle information is preserved better. Secondly, it exploits context information by merging fine-grained shallow feature maps with deep ones full of semantic information, making a better fusion of detailed information and semantic information to achieve a better detection of small faces. Moreover, the HFD exploits prior and multiscale information from the training data to enhance its scale-invariance and adaptability of anchor scales. Compared with the SSH and S3FD methods, the HFD can achieve a better performance in average precision on detecting hard faces as well as a quicker inference. Experiments on the WIDER FACE and FDDB datasets demonstrate the superior performance of our proposed method.

Type: Proceedings paper
Title: Hourglass Face Detector for Hard Face
Event: 2021 International Joint Conference on Neural Networks (IJCNN)
Dates: 18 July 2021 - 22 July 2021
Open access status: An open access version is available from UCL Discovery
DOI: 10.1109/ijcnn52387.2021.9533674
Publisher version: https://doi.org/10.1109/IJCNN52387.2021.9533674
Language: English
Additional information: This version is the author accepted manuscript. For information on re-use, please refer to the publisher's terms and conditions.
UCL classification: UCL
UCL > Provost and Vice Provost Offices > UCL BEAMS
UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Maths and Physical Sciences
UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Maths and Physical Sciences > Dept of Statistical Science
URI: https://discovery-pp.ucl.ac.uk/id/eprint/10135635
Downloads since deposit
2,583Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item