Moodylyser

Emotion detector project that employs Machine Learning

Table of Contents

• About the Project
  • Tech Stack
  • File Structure
• Getting Started
  • Prerequisites
• Usage

• Future Work
• Troubleshooting
• Contributors
• Acknowledgements and Resources
• License

About The Project

#Aim Aim of this project is to predict emotions of a person by analysing a live videofeed of a person. #Description This program anaylses the videofeed of a person frame by frame. It first takes a frame via OpenCV and then crops and resizes the face in the frame to make the frame compatible with the model.Then the landmarks are applied on the face. The model take the cropped frame and classifies it into an emotions out of 5 different emotions it can predict. Refer to our documentation

Tech Stack

This section should list the technologies you used for this project. Leave any add-ons/plugins for the prerequisite section. Here are a few examples.

• Keras
• TensorFlow
• Python
• OpenCV
• Matplotlib
• Numpy
• Google Colab
• Kaggle Notebooks
• Kaggle Dataset(Facial expression challenge)

File Structure

.
├── docs                    # Documentation files (alternatively `doc`)
│   ├── report.pdf          # Project report
│   └── results             # Folder containing screenshots, gifs, videos of results
├── MOODYLYSER2f.ipynb                  # Training program for the Model
├── Moodelld1_5de.h5                  # Pretrained Model with set weights
├── README.md
├── landmarks.py                  # Connects the model to a live videofeed via webcams

Getting Started

Prerequisites

• See SETUP.md if there are plenty of instructions
• The installations provided below are subjective to the machine your are using
• We used [pip install(https://pip.pypa.io/en/stable/)] for the installations. If you don't have pip please follow the following command

 python3 -m pip install -U pip

• List of softwares with version tested on:
  • TensorFlow python3 -m pip install tensorflow
  • Numpy

   python3 -m pip install numpy

  • dlib

   pip install cmake
   pip install dlib

  • Download the Shape predictor file from [here(https://github.com/AKSHAYUBHAT/TensorFace/blob/master/openface/models/dlib/shape_predictor_68_face_landmarks.dat)] and insert it in your current project folder
  • Matplotlib

   python3 -m pip install matplotlib

  • OpenCV

   python3 -m pip install opencv-contrib-python

  • scikit learn

Installation

1. Clone the repo

git clone https://github.com/hashmis79/Moodylyser

Usage

• After cloning the repo transfer the files to your project folder. Open terminal and go to the project folder and run the following commands

cd .../projectfolder
python3 landmarks.py

Results and Demo

Detecting Emotions

Displaying Statistical Data For Emotions

Future Work

• See todo.md for seeing developments of this project

• To Make an emotion detector model
• To connect it to a live feed for live detection
• To give statistical data in the form of graphs
• To increase the accuracy of the model
• To deploy the model in the form of an emotion detector app or site

Troubleshooting

• Common errors while configuring the project

Contributors

• Anushree Sabnis
• Saad Hashmi
• Shivam Pawar
• Vivek Rajput

Acknowledgements and Resources

• SRA VJTI Eklavya 2020
• Refered this for understanding how to use tensorflow
• Refered this course for understanding Convolutional Neural Networks
• Refered towardsdatascience and machinelearningmastery for frequent doubts
  ...

License

Describe your License for your project.