Gender Classification

Training in local

1. Create a directory at root level of this project named dataset.
2. Add 2 sub-directories named men and women containing respective images for men and women.
3. Run train.py for training.
4. Run test.py file for output.

# Trainig
python train.py

# Prediction
python test.py -i <path to input image>

Deploying to Sagemaker

1. Login using AWS SDK in CLI.
2. Run buildpush.sh file located inside container directory.

cd container
./buildpush.sh <image name>

This script accepts one argument for image name. Image is generated using Dockerfile located at /container. After building image, Image will be pushed to Amazon ECR.

Head over to Amazon Sagemaker , create instance and specify to path to image pushed in above script. This image can be used for training and creating endpoints.