This repository contains code for an emotion detection project, which achieves an accuracy of 66%. The project includes four different models, each utilizing a different approach for emotion detection. Below, you will find a detailed description of each model and the corresponding evaluation results.
The project utilizes the "fer2013" dataset from Kaggle, provided by the user "msambare". This dataset consists of facial images labeled with different emotions.
This model is built using a Convolutional Neural Network (CNN) architecture from scratch. The layers used in this model are as follows:
model.add(Conv2D(32, kernel_size=(3, 3), kernel_initializer="glorot_uniform", padding='same', input_shape=(img_width, img_height, 1)))
model.add(Activation('relu'))
model.add(Conv2D(64, kernel_size=(3, 3), padding='same'))
model.add(Activation('relu'))
model.add(BatchNormalization())
model.add(MaxPooling2D(2, 2))
model.add(Dropout(0.25))
model.add(Conv2D(128, kernel_size=(3, 3), padding='same', kernel_regularizer=regularizers.l2(0.01)))
model.add(Activation('relu'))
model.add(Conv2D(256, kernel_size=(3, 3), kernel_regularizer=regularizers.l2(0.01)))
model.add(Activation('relu'))
model.add(BatchNormalization())
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(512, kernel_size=(3, 3), padding='same', kernel_regularizer=regularizers.l2(0.01)))
model.add(Activation('relu'))
model.add(Conv2D(512, kernel_size=(3, 3), padding='same', kernel_regularizer=regularizers.l2(0.01)))
model.add(Activation('relu'))
model.add(BatchNormalization())
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(1024))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(num_classes))
model.add(Activation('softmax'))Evaluation results for this model include:
- AUC_ROC Curve: The area under the Receiver Operating Characteristic (ROC) curve is plotted to measure the model's ability to distinguish between different emotions.
- Confusion Matrix: A confusion matrix is created to visualize the model's performance across different emotions.
- Classification Report: A classification report is generated to provide a detailed analysis of the model's performance, including metrics such as precision, recall, and F1-score for each emotion class.
Similar to Model 1, this model is also built using a CNN architecture from scratch. In addition, augmentation learning techniques are applied to improve the model's performance. The evaluation results for this model include:
- Confusion Matrix: A confusion matrix is plotted to visualize the model's performance across different emotions.
- Classification Report: A classification report is generated to provide a detailed analysis of the model's performance.
- Early Stop Callbacks: Early stop callbacks are implemented to monitor the model's training progress and prevent overfitting.
This model utilizes transfer learning with the VGG16 pre-trained model. The last three layers of the VGG16 model are made trainable, while the rest of the layers are frozen. The evaluation results for this model include:
- Confusion Matrix: A confusion matrix is plotted to visualize the model's performance across different emotions.
- AUC_ROC Curve: The area under the ROC curve is plotted to measure the model's ability to distinguish between different emotions.
In this model, transfer learning is applied using the pre-trained ResNet50 model. The pre-trained weights are used to initialize the model, and the last layer is modified for emotion detection. The evaluation results for this model include:
- Confusion Matrix: A confusion matrix is plotted to visualize the model's performance across different emotions.
- Accuracy: The accuracy of the model is calculated and reported.
To get started with this project, follow the steps below:
- Clone the repository to your local machine:
git clone https://github.com/your-username/emotion-detection.git - Install the required dependencies:
pip install -r requirements.txt - Download the "fer2013" dataset from Kaggle and place it in the appropriate directory.
- Run the models: Refer to the documentation in the respective model directories for instructions on how to run each model.
- View the evaluation results: The evaluation results, including the plotted graphs and generated reports, can be found in the output directories of each model.
The following dependencies are required to run the models:
- Python 3.6 or above
- TensorFlow 2.0 or above
- Keras 2.3.1 or above
- Matplotlib 3.2.1 or above
- Scikit-learn 0.23.1 or above
Contributions to this project are welcome. If you encounter any issues or have suggestions for improvements, please feel free to open an issue or submit a pull request.
This project is licensed under the MIT License. See the LICENSE file for more information.
We would like to acknowledge the following resources that were instrumental in the development of this project:
- msambare/fer2013 Dataset - The dataset used for training and evaluation.
- TensorFlow Documentation - The official TensorFlow documentation, which provided valuable insights into building and training deep learning models.
For any inquiries or further information about this project, please contact anantsrivastava02@gmail.com.
Happy emotion detecting!