Skip to content

satara2358/Deteccion_emociones

Repository files navigation

Emotion_detection_with_CNN

emotion_detection

Packages need to be installed

  • version 3.9.21 of python || env tensor || anaconda
  • pip install numpy
  • pip install matplotlib
  • pip install scikit-learn
  • pip install opencv-python
  • pip install keras
  • pip3 install --upgrade tensorflow ||| optional || pip install tensorflow-cpu
  • pip install pillow
  • pip3 install scipy || opcional
  • conda install -c anaconda scipy || Opcional

Train Emotion detector

  • with all face expression images in the FER2013 Dataset
  • command --> python TranEmotionDetector.py

It will take several hours depends on your processor. (On i7 processor with 16 GB RAM it took me around 4 hours) after Training , you will find the trained model structure and weights are stored in your project directory. emotion_model.json emotion_model.h5

copy these two files create model folder in your project directory and paste it.

run your emotion detection test file

python TestEmotionDetector.py

download FER2013 dataset

Fila 3, columna 3 = 445: El modelo clasificó correctamente 445 imágenes de Happy como Happy.

Fila 3, columna 5 = 376: El modelo confundió 376 imágenes de Happy como Surprise.

Fila 0, columna 3 = 247: El modelo confundió 247 imágenes de Angry como Happy.

❌ ¿Qué te dice esta matriz? El modelo tiende a confundir emociones similares, como:

Angry ↔ Happy

Fear ↔ Sad / Surprise

Happy ↔ Surprise / Neutral

Disgust (clase 1) está muy mal representada → probablemente pocas imágenes o poca capacidad del modelo para distinguirla.

Happy (clase 3) parece tener más aciertos que otras, pero aún así presenta confusiones.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages