Project details:
- Emotion based music recommender system takes the live feed from the camera and predict the mood(emotion) of the user.
- Based on user mood, system recommend the music playlist.
- Playlist are predefined on the database and result will be fetch according to the model predictions.
- User interface will be website created using HTML CSS Bootstrap and JavaScript
- Model may be completely trained from scratch or fine-tunned on pretrained model on architecture such as Mobilenet for low latency improved with haar cascade.
Primary Language: Python
Dataset:
- Fer 2013: https://www.kaggle.com/msambare/fer2013
- Affect net: https://www.kaggle.com/mouadriali/affectnetsample
Deployment: Web based / Website
-
Backend
-
Database (.csv or Sqlite3)
-
Python and library numpy pandas sklearn extensively used
-
Deep Learning model (Tensorflow Keras hyperparameter finetuned model)
-
Model deployment using flask on website
-
Frontend
-
HTML CSS and Bootstrap
-
JavaScript
-
Libraries in use: o Numpy o Pandas o Matplotlib o Cv2 o OS o Random o Tqdm o Tensorflow o Keras o Sklearn o Flask o smtplib