Skip to content

Filestructure

Quang Thien Nguyen edited this page Feb 10, 2021 · 1 revision

Dataset structure

  1. Both training and evaluation datasets have to be stored in the following file structure:
Aaron_Eckhart
    Aaron_Eckhart_001.jpg

Aaron_Guiel
    Aaron_Guiel_0001.jpg

Aarong_Patterson
    Aaron_Patterson_0001.jpg

Aaron_Peirsol
    Aaron_Peirsol_0001.jpg
    Aaron_Peirsol_0002.jpg
    Aaron_Peirsol_0003.jpg
    ...

Labeled Face in the Wild can be used for evaluation but also validation. In order to measure the models accuracy, a pairs.txt file is necessary and also provided in the LFW database.

Clone this wiki locally