Commit Graph

22 Commits

Author SHA1 Message Date
Brandon Amos 678c422894
Improve LFW --lfwPairs error message. 2016-07-10 07:31:05 -04:00
melgor 36cc8ddd72 Correct formating by autopep8 2016-06-29 10:19:08 +02:00
melgor ad078aef9b SpeedUP LFW test 2016-06-23 11:06:28 +02:00
Brandon Amos 18bae7d1ad Update LFW plots. 2016-02-06 23:41:17 -05:00
Brandon Amos a5528889c4 lfw: classify->verify 2016-02-05 20:08:18 -05:00
Brandon Amos 4d855e6704 LFW: Improve error message for missing pairs.txt 2016-01-14 08:40:16 -05:00
Brandon Amos e243655a39 LFW: Improve docs for pairs.txt 2016-01-14 08:35:46 -05:00
Brandon Amos 9d1bd99579 Update copyright notices.
sed -i "/[Cc]opyright.*2015.*Carnegie Mellon University/s/2015/2015-2016/;"  $(git grep '[Cc]opyright.*Carnegie Mellon University'  | sed 's/:.*//' | uniq)
2016-01-13 14:52:52 -05:00
Brandon Amos e8344fbe34 Change lfw interface. 2016-01-12 14:32:54 -05:00
Brandon Amos 46251db1e2 lfw: Output accuracy. 2016-01-11 08:11:37 -05:00
Brandon Amos 434dc1ca9f Fix flake8 warnings. 2016-01-07 18:43:25 -05:00
Brandon Amos 6246fada83 Update code to nn4.v2 for #55. 2016-01-07 18:28:11 -05:00
Brandon Amos dceb6a910b flake8 improvements. 2015-12-24 11:45:00 -05:00
Brandon Amos 58933774c5 Small pep8 fixes. 2015-10-13 07:40:47 -04:00
Brandon Amos ef14b3e4b3 LFW: AUC: Use 3 digits. 2015-10-12 16:55:18 -04:00
Brandon Amos a7de80b529 LFW: Make reference line dotted. Slight alpha on DeepFace line. 2015-10-12 16:52:06 -04:00
Brandon Amos 000f33cab4 lfw experiment: Add argument for the pairs.txt location. 2015-10-10 21:28:41 -04:00
Brandon Amos 69586648e6 Refactor: facenet->openface. 2015-10-10 21:28:41 -04:00
Brandon Amos 1d58e82993 LFW: Show AUC. 2015-10-03 17:16:44 -04:00
Brandon Amos 158ff1bcce LFW: Use 4 decimals. Rename our model. 2015-09-29 18:02:36 -04:00
Brandon Amos 5bffd697cc LFW experiment: Show avg and stdev of accuracies. 2015-09-26 04:41:48 -04:00
Brandon Amos d85ee28f93 Fully implement the LFW experiment.
+ Use ROC curve on 10 folds of test data and average them.
+ Add more ROC curves to the plot for comparison.
+ Compute accuracy from best threshold on the training data.
2015-09-26 04:34:58 -04:00