An app that receives a video as input and returns a chart representing the facial expressions' presence through the video. The video is processed by passing certain frames through a personally developed Convolutional Neural Network, trained on a Kaggle dataset, using keras library in Python. The video and results are then stored in a MongoDB database, being available for accessing later by the user that uploaded the recording. This process is also available for live recordings, the video being analyzed in real-time.

Main functionalities:

• Facial expressions' analysis through a video
• Keeping a history of each user's analysed videos and their results
• Downloading frames for which the CNN is very confident to match a facial expression, perfect for using in other clustering tasks involving facial expressions

Technologies used:

• Python (Streamlit library for the frontend, Keras library for training the AI model)
• FastAPI
• Docker

A demo of the app is available in "Emotional analysis tool demo.mkv"

The project's structure can be seen in "ProjectArchitecture.png"