Matches in SemOpenAlex for { <https://semopenalex.org/work/W4226344269> ?p ?o ?g. }
- W4226344269 endingPage "2474" @default.
- W4226344269 startingPage "2458" @default.
- W4226344269 abstract "This paper addresses the deep face recognition problem under an open-set protocol, where ideal face features are expected to have smaller maximal intra-class distance than minimal inter-class distance under a suitably chosen metric space. To this end, hyperspherical face recognition, as a promising line of research, has attracted increasing attention and gradually become a major focus in face recognition research. As one of the earliest works in hyperspherical face recognition, SphereFace explicitly proposed to learn face embeddings with large inter-class angular margin. However, SphereFace still suffers from severe training instability which limits its application in practice. In order to address this problem, we introduce a unified framework to understand large angular margin in hyperspherical face recognition. Under this framework, we extend the study of SphereFace and propose an improved variant with substantially better training stability - SphereFace-R. Specifically, we propose two novel ways to implement the multiplicative margin, and study SphereFace-R under three different feature normalization schemes (no feature normalization, hard feature normalization and soft feature normalization). We also propose an implementation strategy - characteristic gradient detachment - to stabilize training. Extensive experiments on SphereFace-R show that it is consistently better than or competitive with state-of-the-art methods." @default.
- W4226344269 created "2022-05-05" @default.
- W4226344269 creator A5011157550 @default.
- W4226344269 creator A5016771874 @default.
- W4226344269 creator A5030464101 @default.
- W4226344269 creator A5042278493 @default.
- W4226344269 creator A5070584981 @default.
- W4226344269 date "2023-02-01" @default.
- W4226344269 modified "2023-10-02" @default.
- W4226344269 title "SphereFace Revived: Unifying Hyperspherical Face Recognition." @default.
- W4226344269 cites W1900408618 @default.
- W4226344269 cites W1998808035 @default.
- W4226344269 cites W2076434944 @default.
- W4226344269 cites W2145287260 @default.
- W4226344269 cites W2189774688 @default.
- W4226344269 cites W2325939864 @default.
- W4226344269 cites W2404498690 @default.
- W4226344269 cites W2515770085 @default.
- W4226344269 cites W2605102252 @default.
- W4226344269 cites W2663800299 @default.
- W4226344269 cites W2736633948 @default.
- W4226344269 cites W2798303923 @default.
- W4226344269 cites W2810967842 @default.
- W4226344269 cites W2871667416 @default.
- W4226344269 cites W2904949947 @default.
- W4226344269 cites W2908694412 @default.
- W4226344269 cites W2921310091 @default.
- W4226344269 cites W2939634425 @default.
- W4226344269 cites W2948107928 @default.
- W4226344269 cites W2948697666 @default.
- W4226344269 cites W2949007385 @default.
- W4226344269 cites W2950977493 @default.
- W4226344269 cites W2953214814 @default.
- W4226344269 cites W2955488837 @default.
- W4226344269 cites W2961224374 @default.
- W4226344269 cites W2962898354 @default.
- W4226344269 cites W2962922273 @default.
- W4226344269 cites W2963026686 @default.
- W4226344269 cites W2963350250 @default.
- W4226344269 cites W2963466847 @default.
- W4226344269 cites W2963671154 @default.
- W4226344269 cites W2963814162 @default.
- W4226344269 cites W2963839617 @default.
- W4226344269 cites W2969985801 @default.
- W4226344269 cites W2972986505 @default.
- W4226344269 cites W2983398075 @default.
- W4226344269 cites W2985793077 @default.
- W4226344269 cites W2991234496 @default.
- W4226344269 cites W2991642706 @default.
- W4226344269 cites W3034187833 @default.
- W4226344269 cites W3034303554 @default.
- W4226344269 cites W3034552680 @default.
- W4226344269 cites W3034938706 @default.
- W4226344269 cites W3035364348 @default.
- W4226344269 cites W3035524453 @default.
- W4226344269 cites W3035693354 @default.
- W4226344269 cites W3099206234 @default.
- W4226344269 cites W3101227480 @default.
- W4226344269 cites W3103152812 @default.
- W4226344269 cites W3109225549 @default.
- W4226344269 cites W3109684201 @default.
- W4226344269 cites W3119687891 @default.
- W4226344269 cites W3169129566 @default.
- W4226344269 cites W3176307736 @default.
- W4226344269 cites W3182362926 @default.
- W4226344269 cites W4232886157 @default.
- W4226344269 doi "https://doi.org/10.1109/tpami.2022.3159732" @default.
- W4226344269 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/35294343" @default.
- W4226344269 hasPublicationYear "2023" @default.
- W4226344269 type Work @default.
- W4226344269 citedByCount "6" @default.
- W4226344269 countsByYear W42263442692022 @default.
- W4226344269 countsByYear W42263442692023 @default.
- W4226344269 crossrefType "journal-article" @default.
- W4226344269 hasAuthorship W4226344269A5011157550 @default.
- W4226344269 hasAuthorship W4226344269A5016771874 @default.
- W4226344269 hasAuthorship W4226344269A5030464101 @default.
- W4226344269 hasAuthorship W4226344269A5042278493 @default.
- W4226344269 hasAuthorship W4226344269A5070584981 @default.
- W4226344269 hasConcept C134306372 @default.
- W4226344269 hasConcept C136886441 @default.
- W4226344269 hasConcept C144024400 @default.
- W4226344269 hasConcept C153180895 @default.
- W4226344269 hasConcept C154945302 @default.
- W4226344269 hasConcept C19165224 @default.
- W4226344269 hasConcept C31510193 @default.
- W4226344269 hasConcept C33923547 @default.
- W4226344269 hasConcept C41008148 @default.
- W4226344269 hasConcept C42747912 @default.
- W4226344269 hasConcept C4641261 @default.
- W4226344269 hasConcept C52622490 @default.
- W4226344269 hasConcept C88799230 @default.
- W4226344269 hasConceptScore W4226344269C134306372 @default.
- W4226344269 hasConceptScore W4226344269C136886441 @default.
- W4226344269 hasConceptScore W4226344269C144024400 @default.
- W4226344269 hasConceptScore W4226344269C153180895 @default.
- W4226344269 hasConceptScore W4226344269C154945302 @default.
- W4226344269 hasConceptScore W4226344269C19165224 @default.