EmoLens: Pupil Diameter-based Emotion Classification using CNN and RF Algorithms

This study examines how different emotional states may be categorized using pupil size as a possible indication. The effectiveness of the model in correctly identifying emotions based on pupil size measurements is examined using a CNN-RF (Convolutional Neural Network - Random Forest) model. The study makes use of a dataset that includes measurements of pupil size for various emotional states. Convolutional neural networks and random forests work together to combine their feature extraction and classification skills to create the CNN-RF model. Using the dataset, the model is trained and assessed to determine how well it can categorize emotions. The findings show a considerable fraction of accurately predicted positive events for each emotion, with precision values ranging from 89.47% to 97.35%. The recall values, which vary from 89.11% to 96.30%, show that the model is capable of detecting a sizable proportion of real positive events. The F1 score reaches values in the range of 89.29% to 95.92%, demonstrating a balanced performance in terms of recall and precision. Additionally, the model has high accuracy values between 0.98 and 0.99, indicating that it is generally accurate in predicting emotional states from measures of pupil size. These results imply that pupil size can function as a valid predictor of emotion perception. The CNN-RF model offers a potential method for precisely categorizing emotional states.