Violence-Detection-System

This repository contains the code for a Violence Detection System developed using an LSTM-based deep learning model. The system is designed to classify video clips as either violent or non-violent with high accuracy.

Table of Contents

• Introduction
• Dataset
• Model Architecture
• Training
• Evaluation
• Results
• Usage
• Installation
• Contributing
• License

Introduction

The Violence Detection System leverages the power of deep learning to analyze video content and classify it into violent and non-violent categories. This project utilizes an LSTM (Long Short-Term Memory) network to capture temporal patterns in video sequences, providing accurate classification results.

Dataset

The model was trained and tested on the Kaggle 2000 Videos dataset, which contains a balanced collection of violent and non-violent video clips. The dataset is preprocessed and padded to ensure uniform input shapes for the LSTM model.

Model Architecture

The model consists of:

• An LSTM layer to capture temporal features.
• Dropout layers to prevent overfitting.
• Dense layers with a sigmoid activation function for binary classification.

Training

The model was trained using the Adam optimizer and binary cross-entropy loss function. It was trained for 30 epochs with a batch size of 32, achieving a high level of accuracy.

Evaluation

The model was evaluated on a separate validation set, achieving a 97% accuracy rate. Various evaluation metrics such as accuracy, precision, recall, F1 score, ROC curve, and AUC were used to assess the model's performance.

Results

The system demonstrates a robust performance with the following metrics:

• Accuracy: 97%
• Precision: High
• Recall: High
• F1 Score: High
• AUC: Near 1

Usage

To use the Violence Detection System:

1. Clone this repository.
2. Install the required dependencies.
3. Run the provided scripts to preprocess data, train the model, and evaluate its performance.

Installation

1. Clone the repository:

   git clone https://github.com/your-username/violence-detection-system.git
   cd violence-detection-system

2. Install the required packages:

   pip install -r requirements.txt

3. Download the Kaggle 2000 Videos dataset and place it in the appropriate directory.

4. Run the preprocessing script:

   python preprocess.py

5. Train the model:

   python train.py

6. Evaluate the model:

   python evaluate.py

Contributing

Contributions are welcome! Please open an issue or submit a pull request for any improvements or bug fixes.

License

This project is licensed under the MIT License - see the LICENSE file for details.