Aim is to classify the gender of a person based on his/her photograph. This project assumes binary gender system for simplicity. The model was trained on my laptop's GPU (NVIDIA GTX 1650 4GB).
- main.py - This module is responsible for preparing the dataset and training the model.
- model.h5 - This is the trained model.
Numpy
Tensorflow
Keras
Scikit-learn
Instructions on how to install these libraries can be found extensively on internet.
- main.py - This module’s main aim is to create, prepare and train the model. Internally, also it prepares the dataset which it loads from a specific location in the machine.
Preparing the dataset includes:
- Extracting all the images from a specified location.
- Preprocessing of images which includes:
- Converting all the images to grayscale (to reduce the processing power).
- Resizing all the images to the same dimensions i.e. 80x110 px.
- Creating corresponding output values for each image from the dataset which will be used for training.
Dataset for training has been taken from Kaggle. Thanks to Ashutosh Chauhan for the dataset. You can find it here (270MB).
More information to be added later