Modify: Mood Based Music using Machine Learning and OpenCV

Our face is a vital organ of our body, especially as it plays a significant role in extracting an individual's behaviour and emotional state.If we segregate a music playlist manually, generating a proper playlist based on an individual's emotional features becomes very difficult, time-consuming, labour-intensive, and inefficient.Several algorithms have been developed and proposed to automate this process, but many existing methods remain computationally slow, less accurate, and sometimes even require additional hardware such as sensors.Therefore, our proposed system Modify: Mood-Based Music Using Machine Learning and OpenCV focuses on generating a playlist automatically based on facial expressions, reducing manual effort and processing time while improving system accuracy.The system captures facial expressions using an inbuilt or external camera and employs deep learning-based emotion recognition.Using the FER2013 dataset and Convolutional Neural Networks (CNNs) implemented in TensorFlow/Keras, our model achieved an accuracy of 74.46%.Detected emotions are mapped to predefined or dynamically generated playlists, improving user satisfaction and creating a personalized music experience.Testing was done using both user-dependent (dynamic) and user-independent (static) datasets.Thus, this proposed system provides improved accuracy, faster processing, and lower cost, achieving real-time emotion-based music recommendations using a combination of Machine Learning, OpenCV, and Deep Learning techniques.