Geo-trax: A Comprehensive Framework for Georeferenced Vehicle Trajectory Extraction from Drone Imagery

Geo-trax (GEO-referenced TRAjectory eXtraction) is an end-to-end pipeline for extracting high-accuracy, georeferenced vehicle trajectories from high-altitude, bird’s-eye view drone footage, addressing critical challenges in urban traffic analysis and real-world georeferencing. Designed for quasi-stationary drone monitoring of intersections or road segments, Geo-trax utilizes advanced computer vision and deep learning methods to deliver high-quality data and support scalable, precise traffic studies. This pipeline transforms raw, top-down video data into georeferenced vehicle trajectories in real-world coordinates, enabling detailed analysis of vehicle dynamics and behavior in dense urban environments.

⚠️ Important: If you use the code in your work, please cite the following reference [1]:

Robert Fonod, Haechan Cho, Hwasoo Yeo, Nikolas Geroliminis (2025). Advanced computer vision for extracting georeferenced vehicle trajectories from drone imagery, arXiv preprint arXiv:2411.02136.

(Note: This manuscript shall be replaced by the published version once available.)

The pipeline includes the following key components:

1. Vehicle Detection (✅): Utilizes a pre-trained YOLOv8 model to detect vehicles (cars, buses, trucks, and motorcycles) in the video frames.
2. Vehicle Tracking (✅): Implements a selected tracking algorithm to follow detected vehicles, ensuring robust trajectory data and continuity across frames.
3. Trajectory Stabilization (✅): Corrects for unintentional drone movement by aligning trajectories to a reference frame, using bounding boxes of detected vehicles to enhance stability. Leverages the stabilo 🚀 library to achieve reliable, consistent stabilization.
4. Georeferencing (👷🏼): Maps stabilized trajectories to real-world coordinates using an orthophoto and image registration technique.
5. Dataset Creation (👷🏼): Compiles trajectory and related metadata (e.g., velocity, acceleration, dimension estimates) into a structured dataset.
6. Visualization Tools (✅/👷🏼): Provides tools to visualize the extracted trajectories, overlaying paths on video frames and generating various plots for traffic data analysis.
7. Customization and Configuration (✅): Offers flexible configuration options to adjust the pipeline settings, including detection and tracking parameters, stabilization methods, and visualization modes.

This is a preliminary version of the pipeline with some functionalities not being implemented (👷🏼). Future releases will include more detailed documentation, a more user-friendly interface, and additional functionalities.

• Version 1.0.0

  • Complete georeferencing functionality (Point 4 above).
  • Comprehensive dataset creation with all metadata (Point 5 above).
  • Visualization and plotting tools (Point 6 above).
  • Tools for comparing extracted trajectories with on-board sensor data.
  • Basic documentation and examples covering all core functionalities.

• Version >1.0.0

  • Release tools for (re-)training the detection model.
  • Pre-processing tools for raw video input.
  • Expanded documentation, tutorials (docs folder), and sample examples.
  • List of known limitations, e.g., ffmpeg backend version discrepancies in OpenCV.
  • Comprehensive unit tests for critical functions and end-to-end tests for the entire pipeline.
  • Publishing on PyPI for simplified installation and distribution.

• Version 2.0.0

  • Upgrades to the latest ultralytics (>8.2) and numpy (>2.0) versions.
  • Support for additional tracking algorithms and broader vehicle type recognition.
  • Transition to a modular package layout for enhanced maintainability.
  • Implementation of batch inference and multi-thread processing to improve scalability.
  • Automated testing workflows with GitHub Actions.