<![CDATA[The study investigates the integration of heart rate, gyro, and accelerometer data to enhance emotion recognition across different scenarios like post-movie, post-music, and during music listening. Recognizing the limitations of solely using heart rate data, the research combines gyroscopic and accelerometer data to provide a more comprehensive understanding of emotional responses. Employing machine learning algorithms, notably support vector machines, the aim is to develop robust models for real-time emotion recognition. Conducting rigorous experimental protocols involving motion sensor data from smartwatches, a user study with 50 participants examines emotional responses during activities such as movie-watching and music listening. The dataset includes data from accelerometers, gyroscopes, and heart rate sensors, with additional evaluation metrics to assess the effectiveness of the proposed method in detecting emotional states. The findings demonstrate significant effectiveness of an innovative neural network (NN) method in determining post-activity emotional states, with accuracies ranging from 59.0% to 83.4%, depending on the activity and context. Although NN accuracy is slightly lower compared to other methods like random forest and logistic regression, the differences are not significant, especially when compared to logistic regression. Overall, the research aims to advance emotion recognition technology for applications in human-computer interaction contexts.]]>
