🎯 Battlefield Target Movement Prediction

A geospatial analysis and machine learning system for predicting and visualizing target movements on a battlefield environment.

📋 Project Overview

This project combines geospatial data processing and machine learning to track and predict the movement of targets in a battlefield environment. It includes capabilities for:

• Training prediction models on historical movement data
• Evaluating prediction accuracy against ground truth
• Testing models with various prediction durations
• Visualizing both actual movements and predicted trajectories

---

🧭 Battlefield Geospatial Environment Setup

This guide provides detailed steps to set up a Python virtual environment on Ubuntu for projects involving geospatial processing and machine learning.

📦 System Dependencies

Begin by updating your package index and installing the necessary system-level libraries:

sudo apt update
sudo apt install -y python3-pip python3-venv
sudo apt install -y gdal-bin libgdal-dev
sudo apt install -y libspatialindex-dev
sudo apt install -y libproj-dev proj-data proj-bin
sudo apt install -y libgeos-dev

🧪 Python Virtual Environment

Set up an isolated virtual environment to manage project dependencies:

# Create a virtual environment
python3 -m venv battlefield-env

# Activate the virtual environment
source battlefield-env/bin/activate

🐍 Python Packages Installation

Install required Python libraries, grouped by functionality:

General Utilities:

pip install --upgrade pip
pip install numpy pandas matplotlib requests tqdm

Geospatial Libraries:

Install GDAL with the exact version provided by your system installation:

pip install GDAL==$(gdal-config --version) --no-binary=gdal
pip install rasterio geopandas shapely pyproj

Machine Learning:

pip install scikit-learn
pip install torch

✅ Verification

Test that the key libraries are properly installed:

python -c "import rasterio; import geopandas; import torch; print('All packages imported successfully!')"

🛠️ Troubleshooting GDAL and Rasterio

Should you encounter issues with GDAL or rasterio installations, you may opt to install them via conda:

# Optional: Create a minimal conda environment for geospatial packages
conda create -n geo_deps -c conda-forge gdal rasterio geopandas

You can then use these libraries from the conda environment while continuing development in your Python venv.

📍 Notes

• This setup is tailored for Ubuntu systems. Compatibility with other distributions may vary.
• The use of --no-binary=gdal ensures GDAL builds against system libraries for better integration and compatibility.

---

🚀 Usage

The project offers several modes of operation to support the full workflow from model training to visualization.

🏋️ Training

Train a new prediction model using historical target movement data:

python target_movement_prediction.py --mode train --data_dir data --output_dir models/target_prediction

📊 Evaluation

Evaluate model performance against ground truth data:

python target_movement_prediction.py --mode predict --data_dir data --model_path models/target_prediction/target_predictor_model.pt --output_dir evaluation_results

🧪 Testing

Test the model with a specific prediction duration (e.g., 3600 seconds):

python target_movement_prediction.py --mode test --data_dir data --model_path models/target_prediction/target_predictor_model.pt --prediction_duration 3600 --output_dir test_results

🎨 Visualization

Generate static visualizations of target movements:

python target_movement_prediction.py --mode visualize --data_dir data --output_dir visualizations

🎬 Animation

Create animated visualizations of target movements and predictions:

# Generate animation of actual target movements
python target_movement_animation.py --data_dir data --output_file visualizations/target_animation.mp4 --duration 2

# Generate animation of predicted movements
python target_prediction_animation.py --model_path models/target_prediction/target_predictor_model.pt --data_dir data --output_file visualizations/target_predictions.mp4

# Generate animation with both actual and predicted movements
python target_prediction_animation_with_preds.py --model_path models/target_prediction/target_predictor_model.pt --data_dir data --output_file visualizations/target_predictions.mp4

---

📁 Project Structure

.
├── data/                       # Data directory containing target and terrain data
├── models/                     # Trained prediction models
│   └── target_prediction/      # Target movement prediction models
├── visualization/              # Generated visualizations and animations
├── target_movement_prediction.py  # Main script for training and prediction
├── target_movement_animation.py   # Script for generating movement animations
├── target_prediction_animation.py # Script for animating predictions
├── target_prediction_animation_with_preds.py  # Script for comparison animations
└── README.md                   # This file

---

💡 Features

• Data Processing: Handles various geospatial data formats including CSV, GeoJSON, and raster files
• Machine Learning: Employs deep learning models to predict future target positions
• Time-Based Visualization: Color-coded visualization of movement paths based on timestamp
• Animation: Dynamic visualization of target movements over time
• Prediction Visualization: Compare actual vs. predicted movement paths