Who's Saddest

A web application that captures human face components in photos, detect the emotions on the faces and rank the people based on their sadness.

A WelcomeBackHack2019 Project with Think[Box] in Case Western Reserve University

Getting Started

# Clone the project into a folder
# cd project-folder
# install required
pip install -r requirements.txt
# Django migrations
python manage.py makemigrations
python manage.py migrate
# Run server
python mnaage.py runserver

After running the above steps, visit localhost:8000 to test the application.

Test the program

Test images could be found under test_images/. All images are obtained from the Internet and the owner of this repository does not have the copyright over these images.

Method

Face Detection

Histogram of Oriented Gradients sliding window facial detection.

Emotion Detection

Pre-trained emotion recognition model.

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_1 (Conv2D)            (None, 46, 46, 32)        320       
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 44, 44, 64)        18496     
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 22, 22, 64)        0         
_________________________________________________________________
conv2d_3 (Conv2D)            (None, 20, 20, 128)       73856     
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 10, 10, 128)       0         
_________________________________________________________________
conv2d_4 (Conv2D)            (None, 8, 8, 128)         147584    
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 4, 4, 128)         0         
_________________________________________________________________
conv2d_5 (Conv2D)            (None, 4, 4, 7)           903       
_________________________________________________________________
conv2d_6 (Conv2D)            (None, 1, 1, 7)           791       
_________________________________________________________________
flatten_1 (Flatten)          (None, 7)                 0         
_________________________________________________________________
activation_1 (Activation)    (None, 7)                 0         
=================================================================
Total params: 241,950
Trainable params: 241,950
Non-trainable params: 0
_________________________________________________________________

Tools Used

• Django for the web-backend
• dlib for face detection
• Keras and Tensorflow for emotion detection

Screenshots

• Index page

• Upload images

• Sadness Ranking