Smile Detection Using Datasets and Machine Learning Approach
1 INTRODUCTION
Smile is one of the most beautiful and common facial expression of human being. It expresses various expression. Detecting smile can explain many uses in our everyday life. Even smile is also used as emoji and other representation.
Machine learning is a cutting edge research topic for facial objects detection [1], [2], [3]. These research are based of analyzing datasets of some subject with different facial expression [4]. Though real smiles poses are completely different then these databases.
In this thesis we will try to detect faces from one face dataset. We will train them in such way whether it is a smiling face or non-smiling face. We will take anonymous smiling and non-smiling face and compare them with our trained face dataset. At the end we will use real time smile detection to see if our face is smiling or not.
2 GOAL
Our main goal is to compare an image with our datasets and detect and predict if the image is smiling or not. We also want to find the accuracy of our approach detecting the smile.
3 STATE OF THE ART
There are many but two way is widely used for facial object or expression detection:
1. Geometric approaches.
2. Appearance approaches.
Geometric approaches detect the geometric formation of the face. For facial…show more content… Here we are using an existing dataset, called the “Olivetti faces datasets”. This dataset contains ten different poses of 40 different subjects. So firstly we classify the 400 faces in two categories: smiling and not smiling. For predicting if the extracted face if it is a smiling face or not, we train a support vector classifier in this dataset by using sklearn machine learning library. At the end the classifier is integrated in live loop using OpenCV for capturing a frame from webcam and extract a face and annotate the image with the result of the machine learning