/A-CNN-Based-method-for-Anime-Characters-face-expressions-classification

A CNN Ensemble with three CNN subnets for Anime Character’s face expressions classification with a labeled dataset (angry, fear, happy, sad, surprise, disgust, neutral)

Primary LanguageJupyter Notebook

A-CNN-Based-method-for-Anime-Characters-face-expressions-classification

  • CNN1-drop.ipynb,CNN2-drop.ipynb,CNN3-drop.ipynb are training code of three subnets.
  • CNN1.ipynb,CNN2.ipynb,CNN3.ipynb are training code of three subnets without dropout layer.
  • CNN-mix-drop.ipynb is the ensemble model training code.
  • detect_example.py is a pyqt5 interface for Anime Character Expression Recognization.
  • detect_faces is our dataset on Anime Character Expression Recognization with seven labels.
  • Our trained model parameters can be accessed at Releases.
  • We made a simple image recognition interface using CNN Ensemble model at epoch50 through pyqt5:
  • image