Robust face alignment by dual-attentional spatial-aware capsule networks. (February 2022)

• Record Type: Journal Article
• Authors: Ma, Jinyan; Li, Jing; Du, Bo; Wu, Jia; Wan, Jun; Xiao, Yafu
• Is Part Of: Pattern recognition. Volume 122(2022)
• Publication Date: 2022-02
• Subjects: Face alignment -- Hourglass capsule network -- Adaptively local constrained dynamic routing -- Capsule attention -- Spatial attention; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2021.108297
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: We propose an hourglass capsule network to build a more accurate facial inter-feature spatial constrained model, which enhances the robustness to occlusion and achieves remarkable results. We further improve the original dynamic routing algorithm by adaptively adjust the kernel size to alleviate computational burdens when capturing the landmark spatial positional relationship in the face image. We present a dual-attention mechanism to make the network automatically focus on the more advantageous features and suppress other unrelated ones. Experiment results show that the proposed DSCN achieves excellent performance on challenging benchmark datasets such as 300W, COFW and WFLW. Abstract: Face alignment in-the-wild still faces great challenges due to that i) partial occlusion blurs the inter-features spatial relations of faces and ii) traditional CNN makes the network more difficult to capture the spatial positional relations between landmarks. To address the issues above, we propose a face alignment algorithm named Dual-attentional Spatial-aware Capsule Network (DSCN). Firstly, the spatial-aware module builds a more accurate inter-features spatial constrained model with the hourglass capsule network (HGCaps) as the backbone, which can effectively enhance its robustness against occlusions. Then, two sorts of attention mechanisms, namely capsule attention and spatial attention, are added to the attention-guided module to make the network focus more on the advantageous features and suppress other unrelated ones for more effective feature recalibration. Our method achieves 1.08% failure rate on the COFW dataset, which is much lower than the current state-of-the-art algorithms. The mean error under 300W dataset and WFLW dataset are respectively 3.91% and 5.66%, which shows that DSCN is more robust to occlusion and outperforms state-of-the-art methods in the literature.