student engagement recognition
This is a CNN model to serve as a tool when teachers can't supervise all of students in his/her class.
We imagine there is a camera on the front of each student's desk. They catch a picture every 20 second. These pictures are innput of the CNN model.
In this experiment, we use 'fer_2013.csv' dataset from KAGGLE competitions. We annotate this dataset again, classify these pictures into 3 labels:'engaged','not-engaged' and 'can't define'
Due to the upload limitation, the dataset can't be uploaded because their size.