STAR Loss: Reducing Semantic Ambiguity in Facial Landmark Detection

Recently, deep learning-based facial landmark detection has achieved significant improvement. However, the semantic ambiguity problem degrades detection performance. Specifically, the semantic ambiguity causes inconsistent annotation and negatively affects the model's convergence, leading to worse accuracy and instability prediction. To solve this problem, we propose a Self-adapTive Ambiguity Reduction (STAR) loss by exploiting the properties of semantic ambiguity. We find that semantic ambiguity results in the anisotropic predicted distribution, which inspires us to use predicted distribution to represent semantic ambiguity. Based on this, we design the STAR loss that measures the anisotropism of the predicted distribution. Compared with the standard regression loss, STAR loss is encouraged to be small when the predicted distribution is anisotropic and thus adaptively mitigates the impact of semantic ambiguity. Moreover, we propose two kinds of eigenvalue restriction methods that could avoid both distribution's abnormal change and the model's premature convergence. Finally, the comprehensive experiments demonstrate that STAR loss outperforms the state-of-the-art methods on three benchmarks, i.e., COFW, 300W, and WFLW, with negligible computation overhead. Code is at https://github.com/ZhenglinZhou/STAR.

PDF Abstract CVPR 2023 PDF CVPR 2023 Abstract

Datasets


Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Face Alignment 300W STAR NME_inter-ocular (%, Full) 2.87 # 1
NME_inter-ocular (%, Common) 2.52 # 2
NME_inter-ocular (%, Challenge) 4.32 # 3
NME_inter-pupil (%, Full) 4.03 # 2
NME_inter-pupil (%, Common) 3.5 # 5
NME_inter-pupil (%, Challenge) 6.22 # 2
Face Alignment COFW STAR NME (inter-ocular) 3.21% # 5
NME (inter-pupil) 4.62 # 1
Face Alignment WFLW STAR NME (inter-ocular) 4.02 # 2
AUC@10 (inter-ocular) 60.5 # 4
FR@10 (inter-ocular) 2.32 # 2

Methods


No methods listed for this paper. Add relevant methods here