GSoC 2025 Proposal Submission – “Integrate Fractal ArUco into OpenCV”
Saurav Rijal
Google Summer of Code 2025 Proposal
Project Idea: Integrate Fractal ArUco into OpenCV
Applicant: Saurav Rijal
University: Texas State University
Github: https://github.com/Rizsaurav
LinkedIn: https://www.linkedin.com/in/saurav-rijal-08082a261/
The Idea:
The goal of this project is to introduce Fractal ArUco markers into OpenCV as a new detection pipeline in the existing ArUco module. Fractal markers are a new kind of fiducial marker that are detectable from a longer range and even when there is partial occlusion. Their recursive, layered structure solves some real-world problems with traditional markers—especially when the view of the camera is obstructed or when tracking over a distance is necessary..
The goal is to bring this capability over to OpenCV through an easy-to-work-with API (similar to that of current ArUco capability being used), with clean documentation and a working demo. Integration will be designed with consistency, usability, and accessibility in mind—so users already comfortable using OpenCV's current ArUco tools can immediately start working with Fractal ArUco markers with minimal learning curve.
Why This Matters
OpenCV is already the go-to library for computer vision programming, and its ArUco module is used widely in robotics, augmented reality, and research. However, ArUco's performance deteriorates when markers are far away from the camera or occluded. Fractal markers address this directly. Incorporating them into OpenCV provides this improvement to a wide community of developers and researchers—without depending on third-party libraries.
What I Plan to Do
Design and implement a consistent interface to detect and draw Fractal ArUco markers in OpenCV.
Create a structured marker dictionary system inspired by ArUco’s, but suited to fractal logic.
Include support for estimating pose from visible sub-markers.
Develop a complete walkthrough and sample project showing how it works in real time.
Write up documentation that explains usage, concepts, and limitations clearly for both C++ and Python users.
Plan of Action
Goals:
Deep dive into OpenCV’s ArUco module, Fractal Marker literature, and sample applications. Set up a dev environment, connect with mentors, and finalize implementation strategy.
Define the structure and generation logic for Fractal ArUco markers. Work on the marker dictionary and how fractal IDs will be handled.
Begin implementing the detection logic. Focus on identifying sub-markers and matching them against the hierarchical structure.
Add support for partial detections and pose estimation based on available sub-markers. Internal testing begins.
Integrate the detection logic into a working demo with webcam input. Allow real-time tracking of fractal markers. |
Add Python bindings and finalize the documentation. Polish demo for easy reproducibility. |
Submit final pull requests, complete write-up, gather mentor feedback, and wrap up with a stable, well-tested build.
Related Work and Research Insights
In order to ascertain the feasibility and impact of adding Fractal ArUco markers in OpenCV, I reviewed recent literature and compared marker systems for pose estimation, occlusion resilience, and distant detection.
Romero-Ramirez et al. (2019) introduced Fractal Markers as recursively defined square fiducial marker composition with significantly improved detection range and occlusion resilience compared to traditional ArUco or AprilTag methods. Their hierarchical structure allows for pose estimation even when outer markers are occluded partially or entirely, which is most suitable for dynamic or occluded environments such as drone landing or augmented reality applications in narrow spaces.
As a counterpart to this, Ahmed (2022) compared systematically modern fiducial markers on five points—accuracy, robustness, occlusion, range, and scalability—and reported a lack of standardization in validation techniques. It pointed out that while ArUco and AprilTag continue to be widely used, newer systems like Topotag and LFTag are better under certain conditions, specifically robustness and scalability.
OpenCV reference documentation for ArUco markers (4.12.0-dev) entails detailed technical exploration of marker detection, pose estimation, and dictionary generation. Current implementations, however, remain disadvantaged by range and partial occlusion issues, underlining the imperative to implement advanced solutions like Fractal ArUco markers.
References
Romero-Ramirez, F. J., Muñoz-Salinas, R., & Medina-Carnicer, R. (2019). Fractal Markers: A New Approach for Long-Range Marker Pose Estimation Under Occlusion. IEEE Access. [DOI: 10.1109/ACCESS.2019.2951204]
Ahmed, H. (2022). Fiducial Marker Tracking System: An Approach to Find How Current Markers Are Being Validated. Hochschule Harz. [DOI: 10.13140/RG.2.2.13396.68488]
OpenCV Documentation: Detection of ArUco Markers. OpenCV 4.12.0-dev. https://docs.opencv.org