Paper Title
Face Emotion Recognition with New Auto Generated Emotions Dataset: Emote-2023

Abstract
This paper presents a novel approach to perform facial emotion recognition using transfer learning. The method can be used to automatically detect visual cues of different facial emotions obtained from a variety of multimedia (e.g., movies, TV shows, films, music videos and user generated photos). The proposed approach shows consistent performance and is quite helpful in dealing with the challenge of automatically detecting seven basic human emotions (angry, disgust, fear, happy, sad, neutral and surprise). In addition, a new dataset EMOTE-2023 that includes three new emotions (Contempt, Boredom, Joy) is created using Unreal Engine and the Maya platform. We also extend the approach to classify each facial emotion with more fine-tuned categorical representation. Our work involves performing the transfer learning technique using VGG-16 that allows us to use already pre-trained layers of neural networks on large datasets typically used in computer vision problems. Keywords - Facial Expression Recognition, Transfer Learning, Deep Learning, Convolutional Neural Networks