CNN-on-CelebA-DataSet-using-Tensorflow

Authors: Swati Nair, Jayant Solanki

To implement a convolution neural network to determine whether the person in a portrait image is wearing glasses or not.

We have trained our classification model on Celeb dataset using convolutional neural network and predicted the labels of the test celeb images i.e. whether the person is wearing eyeglasses or not. After training the model using CNN, we achieved an accuracy of approximately 95% on the test images.

code folder consists of below files:

1. Main.py : main code file
2. lib.py : contains helper functions
3. Automate.py: used for hyperparameter tuning

References:

1. Stackoverflow.com,
2. Python, Numpy and TensorFlow documentations,
3. https://cs231n.github.io/convolutional-networks/,
4. Liu, Ziwei; Luo, Ping; Wang, Xiaogang; Tang, Xiaoou. “Deep Learning Face Attributes in the Wild”