This is the authors' abstract. We don't add key points for this paper.
Human face is a very important research subject in computer vision due to its wide application prospect. However, pose, illumination and expression (PIE) variations challenge the robustness offace descriptions. Due to the unique structure and human perception of faces, facial parts are always considered most representative and discriminative in the whole face. In this paper, we propose a novel face representation called Non-Semantic Facial Parts (NSFP). By training a SVM classifier based on identity labels, we automatically find the most discriminative patches on human faces and cluster them to high-level facial parts according to their spatial and appearance correlation. We apply NSF-P to face verification on a public face dataset.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Cao et al. (2015) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: