1 code implementation • 24 May 2024 • Haiyu Wu, Sicong Tian, Aman Bhatta, Jacob Gutierrez, Grace Bezold, Genesis Argueta, Karl Ricanek Jr., Michael C. King, Kevin W. Bowyer
We show that current train and test sets are generally not identity- or even image-disjoint, and that this results in an optimistic bias in the estimated accuracy.
no code implementations • 15 May 2024 • Haiyu Wu, Sicong Tian, Jacob Gutierrez, Aman Bhatta, Kağan Öztürk, Kevin W. Bowyer
In particular, our experiments reveal a surprising degree of identity and image overlap between the LFW family of test sets and the MS1MV2 training set.
no code implementations • 19 Nov 2023 • Haiyu Wu, Sicong Tian, Huayu Li, Kevin W. Bowyer
We explore the potential reasons for this oversight and introduce two pressing challenges to the field: 1) How can we ensure that a model, when trained with data checked for logical consistency, yields predictions that are logically consistent?