Matches in SemOpenAlex for { <https://semopenalex.org/work/W3043609612> ?p ?o ?g. }
- W3043609612 abstract "Most existing public face datasets, such as MS-Celeb-1M and VGGFace2, provide abundant information in both breadth (large number of IDs) and depth (sufficient number of samples) for training. However, in many real-world scenarios of face recognition, the training dataset is limited in depth, i.e. only two face images are available for each ID. $textit{We define this situation as Shallow Face Learning, and find it problematic with existing training methods.}$ Unlike deep face data, the shallow face data lacks intra-class diversity. As such, it can lead to collapse of feature dimension and consequently the learned network can easily suffer from degeneration and over-fitting in the collapsed dimension. In this paper, we aim to address the problem by introducing a novel training method named Semi-Siamese Training (SST). A pair of Semi-Siamese networks constitute the forward propagation structure, and the training loss is computed with an updating gallery queue, conducting effective optimization on shallow training data. Our method is developed without extra-dependency, thus can be flexibly integrated with the existing loss functions and network architectures. Extensive experiments on various benchmarks of face recognition show the proposed method significantly improves the training, not only in shallow face learning, but also for conventional deep face data." @default.
- W3043609612 created "2020-07-23" @default.
- W3043609612 creator A5002140526 @default.
- W3043609612 creator A5017597537 @default.
- W3043609612 creator A5023724461 @default.
- W3043609612 creator A5033870633 @default.
- W3043609612 creator A5042241049 @default.
- W3043609612 creator A5086834120 @default.
- W3043609612 creator A5090649177 @default.
- W3043609612 date "2020-07-16" @default.
- W3043609612 modified "2023-09-23" @default.
- W3043609612 title "Semi-Siamese Training for Shallow Face Learning" @default.
- W3043609612 cites W1509966554 @default.
- W3043609612 cites W1782590233 @default.
- W3043609612 cites W1964414683 @default.
- W3043609612 cites W2115733720 @default.
- W3043609612 cites W2138621090 @default.
- W3043609612 cites W2144172034 @default.
- W3043609612 cites W2145287260 @default.
- W3043609612 cites W2148349024 @default.
- W3043609612 cites W2157364932 @default.
- W3043609612 cites W2194775991 @default.
- W3043609612 cites W2404498690 @default.
- W3043609612 cites W2515770085 @default.
- W3043609612 cites W2520774990 @default.
- W3043609612 cites W2555897561 @default.
- W3043609612 cites W2663800299 @default.
- W3043609612 cites W2737691244 @default.
- W3043609612 cites W2752828042 @default.
- W3043609612 cites W2770121394 @default.
- W3043609612 cites W2771056954 @default.
- W3043609612 cites W2780958074 @default.
- W3043609612 cites W2782415368 @default.
- W3043609612 cites W2798814955 @default.
- W3043609612 cites W2890499605 @default.
- W3043609612 cites W2949007385 @default.
- W3043609612 cites W2961224374 @default.
- W3043609612 cites W2962835968 @default.
- W3043609612 cites W2962895364 @default.
- W3043609612 cites W2962898354 @default.
- W3043609612 cites W2963403266 @default.
- W3043609612 cites W2963420686 @default.
- W3043609612 cites W2963466847 @default.
- W3043609612 cites W2963495494 @default.
- W3043609612 cites W2963671154 @default.
- W3043609612 cites W2963814162 @default.
- W3043609612 cites W2963839617 @default.
- W3043609612 cites W2963976704 @default.
- W3043609612 cites W2964325361 @default.
- W3043609612 cites W2969985801 @default.
- W3043609612 cites W2985817549 @default.
- W3043609612 cites W2998388430 @default.
- W3043609612 cites W2998469040 @default.
- W3043609612 cites W3035524453 @default.
- W3043609612 cites W3099206234 @default.
- W3043609612 cites W3103152812 @default.
- W3043609612 cites W3166898278 @default.
- W3043609612 doi "https://doi.org/10.48550/arxiv.2007.08398" @default.
- W3043609612 hasPublicationYear "2020" @default.
- W3043609612 type Work @default.
- W3043609612 sameAs 3043609612 @default.
- W3043609612 citedByCount "0" @default.
- W3043609612 crossrefType "posted-content" @default.
- W3043609612 hasAuthorship W3043609612A5002140526 @default.
- W3043609612 hasAuthorship W3043609612A5017597537 @default.
- W3043609612 hasAuthorship W3043609612A5023724461 @default.
- W3043609612 hasAuthorship W3043609612A5033870633 @default.
- W3043609612 hasAuthorship W3043609612A5042241049 @default.
- W3043609612 hasAuthorship W3043609612A5086834120 @default.
- W3043609612 hasAuthorship W3043609612A5090649177 @default.
- W3043609612 hasBestOaLocation W30436096121 @default.
- W3043609612 hasConcept C108583219 @default.
- W3043609612 hasConcept C119857082 @default.
- W3043609612 hasConcept C138885662 @default.
- W3043609612 hasConcept C144024400 @default.
- W3043609612 hasConcept C153180895 @default.
- W3043609612 hasConcept C153294291 @default.
- W3043609612 hasConcept C154945302 @default.
- W3043609612 hasConcept C160403385 @default.
- W3043609612 hasConcept C199360897 @default.
- W3043609612 hasConcept C202444582 @default.
- W3043609612 hasConcept C205649164 @default.
- W3043609612 hasConcept C2776401178 @default.
- W3043609612 hasConcept C2777211547 @default.
- W3043609612 hasConcept C2779304628 @default.
- W3043609612 hasConcept C31510193 @default.
- W3043609612 hasConcept C33676613 @default.
- W3043609612 hasConcept C33923547 @default.
- W3043609612 hasConcept C36289849 @default.
- W3043609612 hasConcept C41008148 @default.
- W3043609612 hasConcept C41895202 @default.
- W3043609612 hasConcept C51632099 @default.
- W3043609612 hasConceptScore W3043609612C108583219 @default.
- W3043609612 hasConceptScore W3043609612C119857082 @default.
- W3043609612 hasConceptScore W3043609612C138885662 @default.
- W3043609612 hasConceptScore W3043609612C144024400 @default.
- W3043609612 hasConceptScore W3043609612C153180895 @default.
- W3043609612 hasConceptScore W3043609612C153294291 @default.
- W3043609612 hasConceptScore W3043609612C154945302 @default.
- W3043609612 hasConceptScore W3043609612C160403385 @default.