Facial-Expression-Classification-Deployment

In this project I have used a publicly available dataset from Kaggle. The dataset can be found at (https://www.kaggle.com/c/challenges-in-representation-learning-facial-expression-recognition-challenge/data). This is a facial expression recognition problem from images. The dataset contains a CSV file along with pixel values of 35887 images along with their class labels. There are 28,709 images for training, 3,589 examples for public testing and private test set consists of another 3,589 examples. Facial expression for these images can be of one of these seven classes (0=Angry, 1=Disgust, 2=Fear, 3=Happy, 4=Sad, 5=Surprise, 6=Neutral). To solve this multiclass classification problem I have used three CNN architectures (LeNet-5, a modified version of LeNet and pretrained ResNet-50) with different parameters. The model which generated the best test accuracy was deployed. The deployed model can be accessed through this link https://facial-expression-classifier.herokuapp.com/. For the frontend of the website, I have used HTML5 and CSS. For fetching the result once user uploads an image, I have used Flask.