Gender Detection using Tensorflow

We use the pre-trained Deep Learning Convolutional Neural Network model Inception to identify the gender of person from their photograph.

Refrence:TensorflowInception.

Since Inception in a pre-trained model we try to Transfer Learning to classify the gender with new set of male and female image dataset.

Requirements

Markup : 1. Python 2. Tensorflow

Usage

Prepare Training Dataset

Create folder named training_dataset . Then create folders according to your class-label names(n folders for n classes).Then place all images into the corresponding folder(class-label).

More the variations in images, more the accurate classification.

Start Transfering Learning

$ bash train.sh

Now the Inception model downloads and transferring learning occurs.

Check for Results

Create a folder named test_input that contains all the images to be tested.

python findGender.py

The output will be the predictions for each image in the test data set in the folder test_input.

Results

Check Performance metrics

$ tensorboard --logdir .

Credits

Training images are downloaded here.This repo designed and maintained by Dr Libor Spacek.