Distance (online) Classroom Mood Tracker: Integrating Facial Expression Analysis and Text Analysis

I need a tool to track student moods in distance learning by checking their faces and words in real-time. Here’s what I’m after:

Use Vision Transformers (ViT) to catch emotions from faces in video.
Use OpenCV to spot faces in those videos.
Add Speech-to-Text to turn student audio into text.
Use BERT to analyze if their text is positive, negative, or neutral (sentiment).
Build it with Python.
Make a simple frontend in Python (Flask preferred) to show mood results.
Store data in MySQL.