Overview / Usage

In a class the interation between the students and teacher is very important to achieve a healthy learning process. Many times it happens that students do not pay attention to the class, it may be because of the teacher who is rude or his/her way of teaching is boring or sometimes the teacher remains distracted in the class either on their cellphone or busy doing something else and all these ends up having an unhealthy learning process.
Since our methodology or approach is more focused on the teaching institutes of the rural areas where the number of teacher is less and the number of students is very high so it is difficult for the responsible authorities to assure good quality of learning for the students, even sometimes the teacher is not present in the class and the students start to do whatever they want, therefore here in this project we are trying to solve this problem with computer vision approach which will detect the emotion of the students, teacher, whether the teacher is present in the class or not, whether the students are happy with the teaching or not and it will also detect if the teacher is busy with smartphone and not paying attention to the class and it will create a report based on that.

I will be using Facial Emotion Detection with DNN i.e. Deep Neural Network to detect the emotions of the students as well as of the teacher if they are happy or sad or angry or afraid or disgusted etc.
The smartphone can be detected by Object Detection.

This can also be used to monitor classes if any student is feeling uncomfortable or afraid.

Methodology / Approach

Two different video feed will be taken, one for the teacher side, another one for the student side.

For the student's side,

1. The faces are detected using deep learning face detection to count the the number of students present.
2. The emotions of the faces are detected using Facial Emotion Recognition.

For the teacher's side,

1. The face is detected so that it will distinguish between student and teacher
2. Object recognition is applied to detect the phone of the teacher
3. Teacher's emotion is also tracked

The interaction between teacher and students is found out by comparing the two different video feeds.

Finally the outputs will be shown,

1. Number of students present in the class
2. Teacher : Present or not
3. Overall Emotion of the class
4. Emotion of the teacher
5. Teacher's attentiveness
6. Interaction

We will be using Intel Distribution of OpenVINO Toolkit to accelerate the whole visual inference process.
For the training process Intel DevCloud will be used.
RealSense camera will be used for the teacher's side
Wide angle camera will be used for the student's side to detect all of the students present in the class.

Technologies Used

1. Intel Distribution of OpenVINO Toolkit
2. Intel Optimised Python
3. Intel Powered PC
4. Intel Realsense Camera
5. Intel AI Vision Kit (For wide angle FOV)
6. Intel Movidius Neural Compute Stick