State of the Map Africa 2021

Leveraging AI for Road Mapping in Tanzania
11-19, 10:35–10:55 (UTC), Room 1
Language: English

In Tanzania for the last year our team has been working on digitizing the road network in ten regions including Zanzibar using Artificial Intelligence. Formerly these regions had sparse coverage of OSM data, which made accessibility difficult for the people and businesses across the country. In the end we managed to generate the most comprehensive dataset of roads in Tanzania. Our talk will go into details on the methodology, tools, and technology behind the project.

For the past few years contributing to OpenStreetMap (OSM), the process of creating accurate maps has been a painstaking, time consuming and manual process. Even with the access of satellite imagery and mapping softwares like JOSM and QGIS, the process has still been slow. To help solve this problem, researchers, engineers and developers have developed a new method (Map With AI) that uses deep learning techniques to predict features like buildings and road networks from commercially available high resolution satellite imagery. Map With AI also offers a new set of specialized map-editing services and tools. It includes an editor interface (RapiD) which allows mapping experts to easily review, verify and adjust the map as needed.

As OpenMap Development Tanzania (OMDTZ) we have been mapping several tasks that involve Artificial Intelligence data in road networks. Based on the objectives and the core goals of the project only roads were mapped. Mapping activities were first done remotely and later we sent a team to do ground verification in one of the aforementioned regions (Dodoma). The ground truthing was important because it allowed us to verify the remote work, and also add additional data only obtainable from the ground.

Starting with remote mapping, we used special edition of the OSM Tasking Manager, JOSM software (customized with presets, validation rules, map paint styles) and installed with the MapWithAI plugin (for adding predicted roads to the OSM layer). To ensure high data quality, a professional workflow has been deployed: editing, reviewing, publishing and final checking.

During remote mapping, some features were unclear from the imagery so there was a need to go to the field for verification and add value to the imported road data. Since the imagery we had for the Dodoma region was unclear, it was selected to be the first region for ground truthing. Different road attributes, including street names, the nature of the surface (paved and unpaved), road condition and public transport stops were also mapped to make the mapping more impactful.

In data collection various field mapping tools and mobile applications were deployed to facilitate the task such as OpenDataKit (ODK), Mapillary, and mobile phone cameras. Data collected were uploaded to OSM and to Mapillary.

With respect to the survey conducted in Dodoma the following were the achievements: verification of unclear objects, gaining new skills of data collection to Youth Mappers and community members, new added road attributes to the map, and developed relationships with Institute of Rural Development Planning (IRDP) and the Tanzania Rural and Urban Roads Authority (TARURA).

Despite the achievements mentioned above there are also some challenges encountered during the survey such as accessing some Points of Interest (POI) which were remote and hard to reach, capturing street view images on busy roads was very challenging, and some locations had restricted access.

We plan on completing the remaining regions by the end of the year, and are optimistic about the many ways that the road networks in Tanzania will impact communities and businesses across the country.

Associate Project Manager at OpenMap Development Tanzania (OMDTZ) basing in Tanzania. I joined HOT as a volunteer mapper in March 2015. My first professional engagement with HOT was in a project called Ramani Huria, a project based on Community Mapping for Flood Resilience in Dar es Salaam city. During this project, I experienced data collecting using GPS and field papers at different wards affected by Floods across Dar es salaam. Also experienced in using JOSM editor for cleaning and uploading data to OpenStreetMap as well as remote mapping. I also worked on a project on Malaria Elimination through remote mapping.

Dorice Lucas Mawona,
Mapping Quality Assurance Officer at OMDTZ/HOT Tanzania.

In July 2017, it was my pleasure to join HOT as a volunteer mapper where I was able to contribute to the project of Dar Raman Huria through tracing roads and buildings. The project aimed to create map data that will be potentially useful in helping communities to prepare and escape from floods.

In the mid of February 2018, I was appointed to be a supervisor/ Quality assurance Officer in the digitization team. Within my role, I have been ensuring the quality of the data provided by digitizers through revising their works done every day, creating tasks to be performed, taking statistics of digitized buildings or roads, and reporting to the project manager about the progress of the work. As a supervisor, it is also my responsibility to guide the team and to make them understand the reasons behind why we are digitizing.

Here are the projects I worked on so far, Minigrid Tanzania, Minigrid Zambia, METEOR project, and road coverage in Indonesia, Malaysia, and Tanzania.

Hi! My name is Christina Reuben, Quality Assurance at Open Map Development Tanzania (OMDTZ) Digitization team. I joined OpenStreetMap/OMDTZ team as a graduate with a bachelor of science in environmental science and management. In a short time, I was able to master the mapping processes and now Quality Assurance. My daily activities focus on data validation and quality checks, communication, coordination, training, learning, reporting, and work performance monitoring. Also, I involve with field works for community engagement, data collection, cleaning (analysis), and uploading to OSM. Currently have been working on the road import project in Tanzania using artificial intelligent data to create a road network database, which can be used in disaster response, community service provision, and development initiatives.

I am Neema Justus Alphonce, a Mapping Quality Assurance at Open Map Development Tanzania (OMDTZ). I joined OMDTZ after graduating Bachelor of Science in Land Management and Valuation from Ardhi University Dar es Salaam Tanzania since 2017 as volunteer. I have been working in several projects. In 2018 , I participated in Tanzania Mini grid Mapping project that buildings were only mapped. In 2019 up to date , I am working in Road Import Project as Quality Assurance . My duties based on data quality and quantity, data collection, mapping and providing training to community related to mapping activities and site visit for ground truthing.