Note: This project is in progress, detection is not optimal.
This project aims to help you detect the emotions of different virtual avatars.
- Copy the project and pip install all the requirements:
pip install -r requirements.txt
- Download the Dataset and place it in the project dir with the name "dataset_manga"
- Start the emotion_detector.py file, this will train your model
- When this is done, start the emotion_detector.py file again
- This time pop-up screen will appear and capture the left half of your screen for faces (you can customize the detection size and location in line 142)
- Now you can place different characters' images and detect the emotion of them
- Pressing the q button will close the program
- (optional) You can replace the expression name with the assossiated emotion
I am using:
- CLIP library for obtain embeddings for images
- Anime Face Detector library to detect virtual avatars faces
- Manga Facial Expressions Dataset to fine-tuning the model