There were 33 projects submitted for the #map2020 challenge to build better maps in undermapped regions. Mapping teams joined in from Europe, Asia, and Africa for the chance to demonstrate how street-level imagery can play a role in addressing a humanitarian challenge. Two projects were selected to present at the HOT Summit in Germany.
This spring, Mapillary and Humanitarian OpenStreetMap Team (HOT) joined forces to accelerate map data collection in undermapped regions with the #map2020 campaign. The number of outstanding applications from all over the world blew us away, and after careful deliberation, we have invited two projects, one from Iraq and one from Kampala, to present their work at HOT Summit in September.
Selecting the winning projects from all of the excellent submissions was tough. Ultimately, we were looking for candidates that:
Captured dense street-level images: Projects that captured many images and focused on particular areas were favored.
Edited OpenStreetMap: Imagery is just one step in the process. The important thing is that imagery and derived data were used to add features to OpenStreetMap. We looked at edits made and saved using the #map2020 changeset comment. The quality of these edits and linkage to collected imagery was a key criterion.
Showed ongoing viability: As mentioned at the start of #map2020, a key goal of this initiative was to develop clear lessons that can be shared that show how street-level images can be used to address different humanitarian projects. It is hoped that the selected projects will live on beyond #map2020, and also inspire others to replicate them elsewhere.
Mapping for humanitarian response efforts in Erbil, Iraq
As one of the most active mapping projects in terms of images contributed, the team lead by Mohammed Faisal captured over 103,000 street-level images of Erbil, the capital city of Iraqi Kurdistan. This team experimented with dense image capture in one neighborhood as well as more widespread capture throughout the city, all done with mobile phones and the Mapillary Android app.
The two original team members, Mohammed and Hamza, have backgrounds in humanitarian assistance operations following environmental disasters like earthquakes in northern Iraq. They noticed that because of a lack of adequate and dependable maps, humanitarian agencies were struggling to find routes to reach people in affected areas. Using Mapillary, the two team members began capturing imagery to help during times of crisis. Eventually, the Erbil team grew to become ten mappers and together they improved OpenStreetMap with almost 400 changesets.
“This specific map will be an important tool when a disaster strikes to assess the situation more quickly. The aim is to provide relief workers with the tools to facilitate the decision-making process,” says Mohammed.
The Erbil team’s strategy was to cover all of the main roads of Erbil before focusing their efforts on dense capture in the neighborhood of Andazyaran. Using the OpenStreetMap iD editor, the team focused on adding names and locations of business like pharmacies, supermarkets, and bakeries—all of which can be seen in the Mapillary photo overlay. Next, they used GeoJSON files of automatically generated map data for transformers and manholes to add more detailed point feature information.
The #map2020 team in Erbil added over 103,000 images to Mapillary.
For team lead Mohammed, the next step is to bring his team’s ideas to HOT Summit in September. In addition to presenting their project, Mohammed is looking forward to meeting other mappers at HOT Summit and getting new ideas for how to continue this work.
Highlighting illegal waste disposal in Kampala, Uganda
Led by Henry Sseruwagi, this project came up with one of the best workflows for using street-level imagery to create data on OpenStreetMap. Capturing over 110,000 images, they have shown where illegal dumping of rubbish is occurring in Kampala and added around 50 of those sites to OpenStreetMap.
In Kampala, sanitation is a major problem, especially in the many slums of the city where illegal trash dumping is very prevalent—a lack of proper waste disposal areas and missing trash bins in many areas compounds the situation. Because of this poor sanitation, and congested living conditions, communicable diseases are a major concern. Earlier this year a cholera outbreak spread through Kampala—killing several people and infecting dozens more.
Henry is part of YouthMappers, a student organization in Kampala and it was with three of his fellow YouthMappers that this project idea was born. Their goal was to map the location of illegal dumping sites, so they could present the information to city officials. The core group of four quickly grew to nine mappers, and they covered four areas of Kampala on foot to capture imagery using the Mapillary Android app. The biggest challenge faced by the Kampala team was a lack of reliable internet coverage.
“The internet connection in Uganda is a big problem. We had to let our computers run overnight and then in the morning there would be enough images uploaded,” said Henry Sseruwagi.
The core group of four mappers in Kampala was also responsible for making map edits to OpenStreetMap. Using the Mapillary photo overlay in the iD editor, they manually identified illegal dumping sites to add using the tag
landuse=landfill + informal=yes. Henry mentioned this was the tag that best fit their needs, but the team would have liked to be more specific about each dumping site.
Using the Mapillary photo overlay in the iD editor, the team identified and added illegal dumping sites to OpenStreetMap.
The team’s plan is to use the data they added to OpenStreetMap as a source for the map they present to city officials. As for Henry, he is looking forward to meeting mappers from other disciplines at HOT Summit and learning how they address real-world problems with open data.
The #map2020 campaign has been an effective way to better understand how street-level imagery can be utilized to address a humanitarian challenge. While these tools have been used before, #map2020 was about integrating them from the start with a focus on dense image capture. This allowed projects to download data automatically detected by Mapillary such as road information and the location of objects such as street-lights, catch basins and signage. This led to substantial contributions to OpenStreetMap which you can search with the #map2020 hashtag. Our next efforts will seek to understand which OpenStreetMap tags are most useful for each humanitarian challenge and how data derived from images using computer vision can help to speed up the process.
We were very impressed with the efforts of the various #map2020 projects and hope that all of the groups will continue mapping in their areas. For now, stay tuned for updates from HOT Summit in Heidelberg, Germany.
This blog was also published on Mapillary’s website.