Facial-Attributes-Classification

Facial attributes classification based on MobileNet, a light weight deep neural network using CelebA cropped dataset.

This project is for ENGN8536 in ANU.

Requirement

See requirement.txt for further details.

Dataset

Large-scale CelebFaces Attributes (CelebA) Dataset
CelebA is a large-scale facial images dataset containing 202,599 images with 40 different facial attributes labels.

Model

MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
MobileNets model is on the basis of depthwise separable convolutions, which is a light weight deep neural network.

Experiment

Main enviornment: Pytorch v1.3.1 on Google Colab
Details of our experiment results is in our project report Facial Attribute Classification: A Light Weight Deep Neural Network
Author:
Ziqing Wang: [email protected]
Suikei Wong: [email protected]