EMOJINATOR
- 0 Collaborators
This program converts your facial expression into corresponding facial emoji. ...learn more
Project status: Under Development
HPC, Artificial Intelligence, PC Skills
Groups
Student Developers for AI,
Portland Machine Learning Meetup,
DeepLearning,
Artificial Intelligence Europe,
Movidius™ Neural Compute Group,
Artificial Intelligence India,
Research & Innovation Club of GIT.,
Artificial Intelligence Kenya,
Machine/Deep Learning & Robotics Turkey Meetup,
Intel AI DevCamps,
VIT AI Community,
Deep Learning Brasil
Overview / Usage
This program converts your facial expression into corresponding facial emoji. The model is trained on 450 images of customized faces for each emoji.Right now it supports the following 5 emojis :-
1 - 🙂
2 - 🤫
3 - 😉
4 - 😃
5 - 😑
USAGE-
1- Firstly run face_recorder.py. This will save the customized faces for each emoji into their corresponding labelled folder. Total number of required training images for an emoji is 450 so while executing face_recorder.py the emoji number will automatically iterate on completion of clicking of 450 images.
2- Then run dataset_creator.py to create train and test image dataset.
3- Then run training_model.py to create prediction model.
4- Finally run run.py to convert your real time facial expression into emoji.
5- There is also a file named mask.py which contains the algorithm to mask out required landmarks from the face.
Methodology / Approach
The model is trained on 450 images of customized faces for each emoji. Firstly created mask of face image that contain only eyes,nose and mouth as feature. Then I trained those masked images using CNN for corresponding emoji.
NOTE: The model is trained on the dataset on my face. So it is suggested to recreate dataset by running face_recorder.py.
Technologies Used
1-Keras
2-OpenCV
3-Pickle
4-Dlib
5-imutils
6-shape_predictor_68_face_landmarks.dat