The Digital Humanities hackathon was arranged by University of Helsinki & Aalto University for the second time 16–20 May 2016. The educational event aimed to bring together students and researchers of humanities, social sciences and computer science, for a week of active co-operation in groups under the heading of Digital Humanities.
Exploring the Changing Helsinki
I joined the group who investigated the temporal change of Helsinki through open data and maps provided by Helsinki Region Infoshare, along with cityscape photography by Helsinki City Museum, served through Finna.fi. My fellow investigators were Jussi Kulonpalo (Univ. Helsinki, Social Research/Urban Studies), Timo Laine (Univ. Helsinki, Moral and Social Philosophy), Kari Laine (Aalto University, Computer Networks), Antti Härkönen (Univ. Eastern Finland, General History), Kaisa Sarén (Univ. Helsinki, Bioinformatics), Maija Paavolainen (Univ. Helsinki, Library Sciences) & Rauli Laitinen (Aalto University, Built Environment/ Urban Academy).
We chose Western Pasila as a neighbourhood we wanted to observe. It is the last wooden village that has been torn down to make way for a completely new housing project. At the time, it was also the last remaining freely built working class neighbourhood, dating back to the turn of the 20th century. The area had been under threat of demolition since the 1940’s, which had led to negligence and deterioration of living conditions.
Finna has an impressive collection of openly licensed photographs from the area, taken mainly by photographers of the Helsinki City Museum in a pursuit of documenting the disappearing environment. The photographs do not have exact coordinate location in their metadata, and Timo Laine, who also works for Finna, proposed to use the street name tags to find all Pasila images instead. To get a list of the names, we digitized 2 publications of Helsinki historical street names and used the list to query Finna images.
Reverend Väinö Kantele – Architect Reijo Jallinoja index
Timo came up with a method of evaluating the images by the nature of their keywords. The elements of change were positioned on an axis: old–new, wood–concrete, soft–hard, past–future, freedom–control, community–family unit, Reverend Kantele–Reijo Jallinoja (Reverend Väinö Kantele was a missionary who decided to devote his life to aiding the poor people of Pasila, while Reijo Jallinoja was the architect of the new Western Pasila). Keywords were assigned to one or the other category, and this produced an index number for each image.
Charting the area
I started working towards a street view experiment. First we needed to create a map of the area.
I took a 1969 aerial image from HRI to work with. I used the Wikimaps workflow to do that: I uploaded it to Wikimedia Commons, added the Map template to it and used Wikimaps Warper to georeference it. I was lucky enough to find a few houses that remained from before the development for compositing the then and now images. The file exists also as a readily georeferenced GeoTIFF.
Making a historical map
OpenHistoricalMap is the perfect tool and environment for historical mapping, even though the project is not yet stable. It has all the same tools as OpenStreetMap for creating a map. Features are tagged with start and end dates and that will allow setting and displaying map features on a timeline (in the future).
Turning the map data in OpenHistoricalMap into something that other environments can use is difficult. After trying out many unsupported, broken and obsolete workflows, I found a working solution.
For exporting the data using the OpenStreetMap desktop editor JOSM works nicely. Albin Larsson has written a useful blog post to encompass in making the right settings. OHM’s own export is currently dysfunctional.
Geolocating the images
Historical street view vistas should be easy and painless to create for volunteer contributors. I keep looking for a workflow that is open, easy, collaborative and productive.
I took a test set of 12 images of the same building from different angles over a few years period.
Ajapaik project would be my option for crowdsourcing the locations and directions of the images. This time there was no time for a collaborative effort, so I followed a workflow that was tested in Maptime Copenhagen together with Mapillary.
Geosetter is a great application (Windows only) for setting the location and heading of images, and writing the data in the EXIF metadata section of the image.
It does not seem possible to load a historical map layer into Geosetter, which would be not only useful but essential in order to geolocate historical images.
The advantage of Google Earth is the display of 3D terrain and buildings. I planned to import the converted historical map into SketchUp and define height to the buildings, but I was not able to convert the file to a format SketchUp would understand. It would be possible to use Google Earth to overlay the image onto a 3D view and be able to place it correctly. This workflow should be ported to an open platform.
The street view in Mapillary
After the files were prepared with proper EXIF data, they were ready to be imported into Mapillary. In this test I stopped here. The series of images did not yet resemble a street view experience. There are controls I can use to tie the images correctly together. MapillaryJS is available for further tinkering. I will look into them for the next blog post.
The street view experiments can continue in workshops during the summer. There will be one in Wikimania in Esino Lario, Italy 22–26 June and another one in the Second Swiss Cultural Hackathon in Basel 1–2 July. I will also bring this workflow into discussion in the Beyond the Basics: What Next for Crowdsourcing workshop in the DH2016 conference in Krakow 11–16 July.
Apologies for our group for not presenting all our findings! The presentation will give more insight into that. Thank you for inspiring collaboration during the week!
Thank you for the organizers and participants in other groups for an intensive and energetic week with Digital Humanities hacking!