TUNEX

⇥To build a deploy-able system that detects our emotions at real time using webcam feed and smartly classifies one’s playlist into genres, at last playing a song that suits the current emotion and behavior of the person, as specified by the facial analysis.

Timeline

Date	Task
11-10	Project Assigned and Groups formed on Communication Channel
12-10	Course-1
18-10	Course-2
24-10	Course-3
4-11	Read Papers
11-11	Finalised the approach and the Emotions to detect
14-11	Selection of Dataset and processing
22-11	Face Detection using Harcascade
29-11	Dataset augmented and finalised
22-12	Model Training

Emotions

• Afraid 😨
• Angry 😠
• Disgust 😒
• Happy 😃
• Neutral 😐
• Sad 😭
• Suprised 😮

Dataset

Google Drive link

Demo

Facebook Post

Courses

• Course 1 - Neural Networks and Deep learning
• Course 2 - Improving Deep Neural Networks
• Course 3 - Convolutional Neural Networks

Papers

• Emotion Detections using facial expression recognition and EEG
• Real Time Emotion Recognition System using Facial Expressions and EEG
• Music Recommendation System
• Smart Music Player Integrating facial emotion recognition and music mood recommendation
• Facial Expression Recognition using Facial Landmarks and Random Forest Classifier
• Facial Expression Recognition using Facial Landmark Detection and Feature Extraction via Neural Networks
• Intraface model to extract features

Errors Faced

1. In HaarCascade to detect face, we have to import a xml file which is already present in OpenCV, but when I was doing

face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
#The above xml file was not detected
#maybe because it was not stored in bin of OpenCV as I had installed opencv in miniconda environment

To overcome it, I did this

2. Installing DLib without Pycharm environment: using

   pip install dlib

   give errors because several other packages are required before using this command. To know more refer this

3. While arranging the files in the KDEF dataset, You may encounter a key error on Images 'AF31V.JPG' and 'AM31H.JPG'. The reason for the same is we have seven emotions that are thoroughly documented these two images or least their names do not qualify to our expectations and hence can be removed prior to running this script or this.

4. Visualizing Data in Terminal

5. Visualizing Data in Terminal without Lehar

Made with ❤️ by DRISHTI-AI/ML Team.