Movie Recommendation System Based On Emotion Detection Using Machine Learning Techniques *

The face is a critical perspective in predicting human feelings and moods. More frequently than not human senti-ments are extricated with the utilization of the camera. Various applications are being made based on the location of human sentiments. A few applications of feeling revelation are trade notice suggestion, e-learning, mental clutter, sadness disclosure, criminal conduct discovery, etc. This paper presents a novel real-time emotion-based movie recommendation system that combines computer vision, deep learning, and image processing strategies. The system coordinating OpenCV near DeepFace for effective emotion examination utilizing webcam input, giving clients personalized movie recommendations based on their recognized enthusiastic states. The system commences by using OpenCV to capture real-time webcam feeds and utilizes a Haar-Cascade classifier for facial discovery. The recognized faces are analyzed for prevailing feelings utilizing the DeepFace library, empowering exact feeling distinguishing proof. Within the recommendation stage, the system joins content-based filtering by processing a movie dataset utilizing TF-IDF. Genres and plot keywords serve as features for building the TF-IDF matrix. Cosine similarities between the user's emotion vector and relevant movie genres are then calculated, coming about in a list of personalized movie rec-ommendations. Index Terms-FER, OpenCV, DeepFace, Haar-Cascade Algorithm, Content-Based Filtering, TF-IDF