V. K. Patil - Department of Electronics & Telecommunication Engineering, AISSMS Institute of Information Technology, Pune, India
V. R. Pawar - Department of Electronics & Telecommunication Engineering, Bharti Vidyapeeth College of Engineering, Pune, India
S. P. Kulkarni - Department of Electronics & Telecommunication Engineering, AISSMS Institute of Information Technology, Pune, India
T. A. Mehta - Department of Electronics & Telecommunication Engineering, AISSMS Institute of Information Technology, Pune, India
N. R. Khare - Department of Electronics & Telecommunication Engineering, AISSMS Institute of Information Technology, Pune, India

Emotions are the accelerators of human intellect and innovation and creativity, so the ability to recognize emotions is in high demand. Real-time hardware has hurdles of Noise and hardware factors as compared to simulations. An electrocardiogram (ECG) sensor (AD۸۲۳۲), a temperature sensor (LM۳۵), and a signal processing circuit is  hardware of the proposed real-time emotion identification. The RR intervals are calculated from the ECG data. Emotions prediction using machine learning makes use of RR intervals and body temperature as features. One of the four emotions (namely ۱. Happy ۲. Stressed ۳. Neutral ۴. Sad.) is visualized at the serial port of the processor by using WESAD benchmark dataset and the HRV, serial, and pickle libraries.This article's innovation factors are (۱) Use of ECG for emotion detection rather than disease detection with Emotion induction method, RR interval capturing and design of RR interval GUI for real time capture of temperature and ECG (۲) Display of current emotion on Arduino serial port. (۳) Measurement of Class performance using F۱ score, macro average, and weighted average instead of general term accuracy. (۴) Use of the probability based  Navies Bayes as compared to traditional KNN, SVM, Random Forest nethods  (۵) Class wise performance for example Navies Bayes' specificity or accuracy is lower than SVM's (۰.۹۶), but its recall or sensitivity is higher (۰.۹۷) vs. (۰.۹۴) for stress.In this article, we presented performance parameters in terms of interactive computations, tabular form and graphical display.

Emotion Recognition, Machine Learning, Support Vector Machine, Naive Bayes, Electrocardiogram, heart rate variability