You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Overview no método enquanto organizava os papers no Zotero...
85% em ambos KinFaceW-I e II.
76.80% no FIW.
Geralmente comparamos a média, mas comparar a performance nos tipos de parentesco pode ser útil. Um método específico pode ter algum componente interessante para um ou mais tipos. Ver comparativo abaixo. Resultados da Hierarchical Representation Learning for Kinship Verification #66 reportados aqui.
Kinship
Method
B-B
S-S
SIBS
F-D
M-D
F-S
M-S
GF-GD
GF-GS
GM-GD
GM-GS
Average
Siblings
AIAF + IFW
73.86
85.57
77.60
76.80
FaCoRNet
82.48
84.48
81.66
81.87
Parent-Child
AIAF + IFW
79.13
78.22
76.16
86.59
FaCoRNet
79.36
81.53
84.85
79.90
Grandparent-Child
AIAF + IFW
69.37
69.36
70.53
73.88
FaCoRNet
78.78
68.16
73.23
56.42
Essa estratégia lembra bem o DisVAE (#40). Inclusive, seria interessante para o trabalho de Lucas o mecanismo Adversarial Canonical Correlation Regularizer (ACCR).
No description provided.
The text was updated successfully, but these errors were encountered: