Enhancing global mapping initiatives by developing an open-source, AI-assisted tool, that enables efficient creation, training, and deployment of models for feature extraction from aerial and drone imagery, improving the quality and accessibility of open data. In this 8-week challenge, you will join a collaborative team of 50 AI engineers from all around the world.

The problem

Updating and enhancing global maps with detailed geographic and infrastructural features is a process fraught with challenges, primarily due to its reliance on time-consuming and resource-intensive manual methods. This traditional approach to mapping, which involves manually identifying and cataloging changes in the landscape, is not only slow but also prone to inaccuracies. As a result, the availability of up-to-date and accurate geographic data is often limited, creating significant hurdles for effective decision-making and response strategies across various critical sectors, including disaster management, urban development, and environmental protection.

The impact of these challenges is profound. In disaster response scenarios, for instance, the absence of timely and precise map updates can severely impede rescue and relief operations, potentially exacerbating the vulnerability of affected populations. For urban planning and development initiatives, reliance on outdated maps can lead to inefficient resource allocation, planning errors, and overlooked opportunities for sustainable development. Similarly, environmental monitoring and conservation efforts suffer when changes in natural resources and land use cannot be quickly and accurately mapped, hindering actions aimed at protecting ecosystems and biodiversity.

In light of these challenges, there is a pressing need for innovative solutions that can accelerate the mapping process, enhance the accuracy of the data collected, and ensure that comprehensive, up-to-date geographic information is readily accessible. This is where the collaborative initiative between the Humanitarian OpenStreetMap Team (HOT) and Omdena to develop an open-source, AI-assisted mapping tool comes into action. The project seeks to address a significant challenge in the field of global mapping: the labor-intensive and slow process of manually updating maps with detailed geographic and infrastructural features. By leveraging AI for feature extraction from aerial imagery, this project aims to transform the global mapping landscape, making it possible to rapidly update maps with high levels of precision and significantly improving accessibility to critical geographic data.

The goals

The ultimate goal of this project is to enhance the capabilities of the AI-assisted mapping tool, enabling end users to efficiently create, train, and deploy AI models for feature extraction from aerial and drone imagery. This initiative aims to significantly improve the quality and accessibility of open data, supporting a wide range of applications from urban planning to environmental monitoring.

The main goals of this challenge are:

• Development of AI Models for Advanced Feature Extraction: The project will focus on expanding the AI models’ capabilities, testing various models such as Segment Anything (SAM), FastSAM, GroundingDINO (DINO + SAM), and Yolov8 + SAM, among others suggested by AI challenge collaborators. This involves leveraging cutting-edge machine learning algorithms and techniques to enhance the tool’s ability to accurately identify and extract features from complex imagery.
• Creation of Comprehensive Training Datasets: Developing training datasets will be a critical task, utilizing Open Aerial Map imagery and OpenStreetMap (OSM) data. This goal ensures that the AI models are trained on high-quality, diverse datasets that reflect the real-world complexity of geographic features.
• Evaluation of Model Performance: The project will rigorously evaluate the performance of each model based on precision, recall, and other relevant metrics. This systematic assessment aims to identify the most effective models for feature extraction, ensuring that the tool meets high standards of accuracy and reliability.
• Ensuring Open Access and Collaboration: All libraries and dependencies used in the project will be free and open source, and the datasets will be public, provided by the partner. This approach promotes collaboration within the community and ensures that the project’s outcomes are accessible to a wide audience.

By achieving the above goals, this project aims to deliver a powerful, user-friendly tool that democratizes access to advanced AI capabilities for feature extraction from aerial and drone imagery. This initiative not only advances the state of open mapping initiatives but also empowers users across various sectors to leverage AI for creating more accurate, up-to-date maps, thereby facilitating informed decision-making and innovation in addressing global challenges.

Why join? The uniqueness of Omdena AI Innovation Challenges

A collaborative experience you never had in your working life! For the next eight weeks, you will build AI solutions to make a real-world impact and go through an entire data science project lifecycle. This covers problem scoping, data collection, and preparation, as well as modeling for deployment.

And the best part is that you will join a global and collaborative team of changemakers. Omdena AI Challenges are not a competition or hackathon but a real-world project that will take your experience of what is possible through collaboration to a new level.

Find more information on how an Omdena project works