Profile Picture Classifier Based on Ethnicity

This Web App uses the EfficientNet_Lite0 model to build an API that predicts the ethnicity of Profile Pictures.

Steps

The scrape_images_of_ethnicity.ipynb notebook was executed locally to scrape two hundred and twenty (220) images of asian and black americans from google. The scrapper only scrapped for 2 ethnic classes, namely:

• Asian Americans and
• Black Americans

These classes were selected on the basis of simplicity and can be improved upon. The two selected ethnic classes had 110 images each in a class-separated folder. 100 images from each class was used for training while 10 was used for validation.

Link to the dataset can be found here

DataLoader

The classify_profile_picture.ipynb notebook was executed in google colab for efficeincy and speed. The images from the folders were loaded using the DataLoader function from tflite_model_maker.image_classifier class. The DataLoader function together with the from_folder method was used to load images from subdirectories, identifying the subdirectory names as the class labels.

Model Training

The default EfficientNet_Lite0 model from tflite_model_maker.image_classifier.image_classifier.create class was used to train the images for 50 epochs achieving an accuracy score of 90.0% on the validation dataset. A very good score considering the very small volume of data at the model's disposal.
The tflite_model_maker model is stored in the model_dir folder as model.tflite.

Running the service...

• Clone this repo
• Create an isolated python virtual environment and install dependencies in the requirements.txt file
• Go to the project directory and run python3 api.py
• Click the link to open localhost
• On the homepage of the web app click on Choose File to select an image
• Select a toyota image of either an asian or black american from your local machine
• Click on the Predict button to the right
• Wait a few seconds for the system to process and predict the ethnicity of the profile picture.
• Viola!!! Here you have your prediction.

Snapshot of Web App

The home page

The prediction Page

Further Steps

• Create a more sophisticated web interface
• scrape more image dataset to create an even better model
• Include more ethnic classes