I love working with ML models and have written various articles explaining some of them. I developed an emotion detection model using CNN as part of my academic project.
We used MMI datasets for testing our models and tried different ML models including GNB, LR and SVM.
Overall for videos, CNN performed much better than the traditional models as there were no features given and feature extraction was done by the model itself.
For the image dataset, we ran the model on given features of the image and SVM performed the best in this scenario.