search

dsp-uga / purah-p2

MIT License
0 stars 0 forks source link

Play with Dataset #2

Closed nathanWynn closed 3 years ago

nathanWynn commented 3 years ago

Need to familiarize ourselves with the provided dataset. Might be useful to write a "hello world" python file to interface with the GS bucket and poke around.

mashihan123 commented 3 years ago

There are 39 pre-extracted features in the CSV files, which are used for training neural networks for the classification task. The last column is the label for each detected face, either 1 or 0. Therefore, this is a binary classification problem. There are over 40 GB images in the GCP bucket which are used for feature extraction. The training list contains over 654,000 images.