Fight Action Prediction

This project focuses on predicting fight actions using deep learning models. We classify various fight-related actions using models like ResNet152 and EfficientNetB7. The actions being predicted include:

• 0: Hit
• 1: Kick
• 2: Punch
• 3: Push
• 4: Ride Horse
• 5: Shoot Gun
• 6: Stand
• 7: Wave

Models

We trained and fine-tuned the following models:

• ResNet152: Fine-tuned on the action dataset.
• EfficientNetB7: Another deep learning model used for comparison.

Dataset

The dataset for training includes labeled images corresponding to the actions mentioned above. Ensure your dataset follows the directory structure:

data/
│
├── train/
│   └── <class_name>/
│       └── image1.jpg
├── test/
    └── <class_name>/
        └── image2.jpg

Each class folder contains relevant images for that action.

Installation

Clone the repository and install the dependencies:

git clone https://github.com/alihassanml/Fight-Prediction-Project.git
cd Fight-Prediction-Project
pip install -r requirements.txt

Training

To train the models, run the following command:

python train.py

This will start training the model on the dataset located in the data/train directory.

Streamlit Application

You can interact with the trained models using a web interface powered by Streamlit. To run the app:

streamlit run streamlit_app.py

Upload an image, select the model you want to use, and get predictions for the action being performed in the image.

Results

After training, models will be saved as .h5 files, and the model performance can be assessed using validation data and metrics such as accuracy, confusion matrix, and classification report.

Fine-Tuning

Fine-tuning is done by unfreezing the last few layers of the pre-trained models and retraining them with a smaller learning rate to improve accuracy.

License

This project is licensed under the MIT License.