So far this has been the century of “my do it.” You know “my do it” — it’s what 2-yr-olds say after they’ve watched you do something, like pour detergent into a washing machine, tie a shoe or press buttons in an elevator, and now it’s their turn. They can’t yet perform the act well on their own, but if you try to step in and help? Listen to their voice as it reinforces their singular determination: “NO! My do it!” This is the way a child learns, contributes, expresses her independence and participates in the world around her.
Different context, different ages, but the motivational seeds behind “my do it” can be seen within the processes of user-generated content, crowdsourcing, citizen science, volunteered geographic information and related people-powered activities. A decade ago, it was a defining characteristic of Web 2.0. This isn’t just a trend or fad; we’re in a different paradigm for how the world works. Not only are data opportunities simply too vast to manage in traditional ways, but our eyes have been opened to the advantages of being directly involved in data gathering, for example: to be able to affect and control the timing, the geographic extent or coverage, the resolution, the granularity, the accuracy. It’s a matter of control, and once the learner knows how to do something on his or her own, and gets satisfaction and benefits from doing so, why would he or she want to revert to being passive in the process?
Plus, we want to share that data. Not only do we have the pride of having gathered it on our own, but we seek mechanisms for sharing it in as universal a way as possible. Data by the people, for the people.
One new company filling a niche or two in the world of geospatial user-generated content is Mapillary. Founded in early 2014 by CEO Jan Erik Solem, Mapillary allows users to generate, contribute and view geotagged street-level photos from around the world. It’s the people’s version of Google’s Street View.
If Google is already providing us with Street View, why duplicate those efforts? To provide images under a different agenda from Google’s, with higher temporal resolution, higher image resolution and an expanded geographic extent. With few exceptions, Google has necessarily focused on the more heavily trafficked, grander veins and arteries of the human transportation network. Mapillary is deliberately targeting the smaller roads, paths and trails that play a significant role in the human experience, though they may not be as prominent or visible as other parts of the network. You know: Mapillary, capillary.
Mapillary capitalizes on the fact that placing a photo on the surface of the earth based on its point of origin can be done automatically, and most accurately, if the digital image’s file is tagged with latitude and longitude coordinates in its Exif metadata when it is originally taken. These days many more pictures are taken via a smartphone than a GPS-enabled camera, and as long as the location services are enabled for the phone’s camera app, the image is geotagged as it is snapped. Manually “placing” a photo in its place on a digital map after the fact can only approximate that information.
Moreover, Mapillary’s focus is not about the viewing of individual photos in their geographic places; Flickr and Panoramio have that functionality well covered. While already-geotagged photos can be uploaded to the Mapillary website, its smartphone app facilitates the automatic capture of multiple photos in succession, while you’re walking, for example, so the resulting geotagged collection becomes a recording of a journey more than a single snapshot of a place. Last fall Mapillary held its first contest to recognize excellent walks. The winning entry, submitted by user didymops, is a foray into a snowy Canadian forest. Though it seems a little like choppy video footage, the absence of an audio stream also evokes the natural silence of that place and time.
Why not use video itself to share such walks? That’s possible, and YouTube is flooded with them. But metadata for geotagging video formats are not as standardized as Exif, and such geotagging is central to Mapillary’s objectives.
The rate at which geotagged pictures are taken by the Mapillary app can be modified, an important feature since they have also announced a partnership with the action camera maker GoPro. By connecting via WiFi to a nearby GoPro4 camera, the phone-based Mapillary app will be able to record location coordinates of images that are being snapped – every second or so – with the GoPro camera, rather than the smartphone’s camera. Of course it won’t yet work while underwater images are being taken, though that’s on my own particular wish list for the developers!
However, CEO Solem explained that maximizing temporal resolution is not as important to them as ensuring that the images themselves have high pictoral resolution. This supports another of their objectives: to enable the georeferenced images to be used as data sources for generating 3D models of objects in the photos themselves. Accomplishing this harkens back to the days when Microsoft launched Photosynth and stitched together many single images, based on object identification and matching in the images, enabling a user to have smooth, transition-free, video-like visual experiences. By taking this intent in a slightly different direction (no pun intended), computer scientists have also developed algorithms that rely on the resulting digital point-clouds to recreate urban landmarks such as the Roman Coliseum. A similar recent project at the University of North Carolina has used the bounty of Flickr images and leveraged computer processing speeds to illustrate how one could reconstruct a digital 3D world in six days.
Mapillary is making this type of activity more accessible to the public by having the heavy lifting of processing take place in the Cloud. Their object identification process is being applied to traffic signs and users can contribute their superior vision to this process via a traffic sign game (login required). To further promote their goal of 3D model building, they have made it possible to view digital point-clouds from uploaded imagery, though the visual result is likely to be underwhelming and confusing for the many places with too few photos. Only when a large enough number of images have been contributed and processed can the resulting model be used to measure dimensions and distances and other such tasks. Still, the potential is clear, and for this reason Mapillary has attracted the attention of organizations such as the World Bank, which is now working with the technology to process images from Dar es Salaam as they address chronic flooding problems.
The “my do it” metaphor between toddlers and adults isn’t a perfect one for the transition to crowd-sourced geotechnology platforms. For one, it’s missing the role of companies who purposefully strive to help newbies leap from novices to practitioners within a short amount of time. It also miscasts large technology companies as ones with only “Stand back, I’m an expert in this” attitudes, losing patience with two-year-olds trying to reach the elevator buttons and disinterested in helping them achieve. In contrast, partnerships like the one between Mapillary and Esri are just as likely in today’s world. That’s the much more realistic image of future business growth and development. Still, the insistent eagerness of early learners will continue to be a driving force in the technology world.