Skip to content

Latest commit

 

History

History
96 lines (75 loc) · 5.14 KB

README.md

File metadata and controls

96 lines (75 loc) · 5.14 KB
Raw

Music Recommendation based on Facial Emotion Recognition

Mood Detection model can detect face from any image and then it can predict the emotion from that face. We can do it from both still images and videos. After predicting the emotion from face our recommender system take the predicted emotion as input and generate recommendation by processing a Spotify dataset from a kaggle contest. We predicted the music mood from a model trained with data_moods.csv. The recommender system will generate top 40 songs to recommend for a spotify playlist.

Mood Detection

Mood Detection model will predict one of the emotion among 7 emotions listed below-

  • Happy
  • Sad
  • Angry
  • Disgust
  • Surprise
  • Neutral
  • Fear

Music Mood Prediction

Every songs in the main dataset in the Datasets.7z folder predicted to one of the mood among 4 moods listed below-

  • Happy
  • Sad
  • Energetic
  • Calm

By using a music mood classifier model we predicted each songs mood in our intermediate dataset kaggleSpotifyMusicMood in the Dataset.7z folder.

Music Recommendation

Our main project file is music_recommender.ipynb file. This recommendation system is using content based filtering. We follow these steps to recommend music-

  • Dataset Pre-processing
  • Feature Engineering
  • Connect to Spotify API
  • Create Playlist Vector
  • Generate Recommendation using cosine similarity

So according to this project, we will take an image of an user and predict emotion using Emotion Detection model. By prioritizing the songs from our main dataset kaggleSpotifyMoodFinal.csv with music mood comparing with different face emotion, this system will generate top 40 songs to recommend for a particular spotify playlist.

Spotify Dataset link: https://www.kaggle.com/yamaerenay/spotify-dataset-19212020-160k-tracks (Spotify Dataset 1922-2021, ~600k Tracks)
Dataset for mood Classifier link: https://github.com/cristobalvch/Spotify-Machine-Learning/tree/master/data (data_moods.csv)

Project Description:

The emotion recognition model is trained on FER 2013 dataset. It can detect 7 emotions. The project works by getting live video feed from web cam, pass it through the model to get a prediction of emotion. Then according to the emotion predicted, the app will fetch playlist of songs from Spotify through spotipy wrapper and recommend the songs by displaying them on the screen.

Features:

  • Real time expression detection and song recommendations.
  • Playlists fetched from Spotify using API.
  • Neumorphism UI for website.

Running the app:

Flask:

  • Run pip install -r requirements.txt to install all dependencies.
  • In Spotipy.py enter your credentials generated by your Spotify Developer account in 'auth_manager'. Note: - This is only required if you want to update recommendation playlists. Also uncomment import statement in 'camera.py'.
  • Run python app.py and give camera permission if asked.

Tech Stack:

  • Keras
  • Tensorflow
  • Spotipy
  • Tkinter (For testing)
  • Flask

Dataset:

The dataset used for this project is the famous FER2013 dataset. Models trained on this dataset can classify 7 emotions. The dataset can be found here.

Note that the dataset is highly imbalanced with happy class having maxiumum representation. This might be a factor resulting in okaysish accuracy after training.

Model Architecture:

  • The model architecture is a sequential model consisting of Conv2d, Maxpool2d, Dropout and Dense layers:
  1. Conv2D layers throughout the model have different filter size from 32 to 128, all with activation 'relu'
  2. Pooling layers have pool size (2,2)
  3. Dropout is set to 0.25 as anything above results in poor performance
  4. Final Dense layer has 'softmax' activation for classifying 7 emotions
  • Used 'categorical_crossentropy' for loss with 'Adam' optimizer with 'accuracy' metric

Note:- Tried Implementing various other models like VGG16 but accuracy was far too low. This model architecture gives good enough accuracy. A bit more tinkering with hyper parameters might lead to a better accuracy

Image Processing and Training:

  • The images were normalised, resized to (48,48) and converted to grayscale in batches of 64 with help of 'ImageDataGenerator' in Keras API.
  • Training took around 13 hours locally for 75 epochs with an accuracy of ~66 %

Current condition:

The entire project works perfectly fine. Live detection gives good frame rates due to multithreading.

Project Components:

  • Spotipy is a module for establishing connection to and getting tracks from Spotify using Spotipy wrapper.
  • haarcascade is for face detection.
  • camera.py is the module for video streaming, frame capturing, prediction and recommendation which are passed to main.py.
  • main.py is the main flask application file.
  • index.html in 'templates' directory is the web page for the application. Basic HTML and CSS.
  • utils.py is an utility module for video streaming of web camera with threads to enable real time detection.
  • train.py is the script for image processing and training the model.