Image mapping parties are not only organised to capture images. They also provide a great opportunity to teach people how to use street-level imagery to improve Open Street Map.
I’m looking forward to some image mapping parties coming up in the next few weeks.
The first in the New Forest (Sunday 13th September 2020), run in conjunction with Nick Whitelegg the creator of OpenTrailView.
The second is with the Bexhill OSM project, who have already used a GoPro MAX to map a large part of their area.
Let’s hope the great British weather improves by then (I’m sitting writing this post on a mild, overcast 3 day-weekend).
Image mapping parties are not only organised to capture images. They also provide a great opportunity to teach people – anyone, and everyone – how to use street-level imagery to improve Open Street Map. Most importantly they’re a great opportunity to meet up with fellow hobbyist mappers.
In this post, I wanted to outline some of the things we cover before, during and after image mapping parties. This post is not designed to be a logistical guide, more a pointer to helpful resources to ensure attendees get the most out of the day. If you do want a step-by-step guide to organisation, the OSM How-to wiki for traditional (non-image capture) mapping parties will prove incredibly useful for event organisers too.
The aim; to inspire you to host an image mapping party in your local area.
Before the mapping party
Many people think they need lots of expensive equipment for a successful image mapping party. This is simply not the case.
If you can get your hands on a 360 camera and set it up like a Trek Pack, that’s great. But it’s not essential.
I am a big believer in something is better than nothing for image mapping parties.
A single lens action camera takes great images. OK, they’re not 360’s but it’s something.
Failing that, most people have smart phones with software to create panoramas. Or even just 2D photos.
Mapillary is mostly comprised of forward facing dash-cam photos that have proved valuable for street-level object recognition software training.
Having a series of photos every 10 meters taken by a smartphone is also incredibly useful for those editing Open Street Map.
Even with professional equipment, it will often take a couple of runs to account for mistakes and issues. Remember, you can always recapture at a later date… like you needed an excuse for another party.
Arguably, it’s more important people understand how the imagery will be used post-capture.
By getting to grips with basic camera and mapping concepts, people will learn how to use the equipment to take useful photos for improving OSM.
- Yaw, pitch and roll
- GPS 101
- Photo and video metadata
- FPS, Bit rate, Compression and 360 Video Quality
Attendees don’t need to be an expert on any of these topics, but simply sharing them a few days before the event, and again on the morning of the event, can lead to a lot of avoidable mistakes by helping them understand both what they need to do, and why.
Show, don’t tell is an important concept.
In addition to the links above, I also like to get people to understand the value of the imagery and how it will be used (by them and others) post-capture.
For this, I share this post, Using Street-Level Imagery to Improve the Map, and ask attendees to look at Mapillary imagery around their home location or where the mapping party will take place in the OSM iD Editor.
Remember, not everyone will be familiar with OpenStreetMap. In fact, this might be their first experience with it. In such cases, users can get a feel for how OSM works and where the photos they take during the mapping party will be used.
Which brings me nicely onto my next point; have a plan of where you want to map.
By doing so, you can plan for teams to capture imagery most effectively and avoid people standing around on the day.
Don’t make it too systematic. The exercise is meant to be fun and things rarely run to plan. Just make sure to have a clear outline of:
- What needs to be captured and by what team (what roads? Pathways? Need to capture crossing points?)?
- How should people should capture images (if using phones, should users do both sides of streets in opposite directions? Is this needed for 360 cameras? How should cameras be positioned / mounted?)?
- What resolution imagery should be captured at (how far should images should be spaced? What camera settings are needed?)?
- Where and when should everyone meet-up at then end of the day (I suggest a nice pub!)?
Unlike traditional OSM mapping parties where data is uploaded once mapping is complete, it usually makes sense to handle the upload and processing of imagery once everyone is at home due to the amount of data that needs to be uploaded (most free WiFi networks are not particularly fast).
Who uploads imagery will all depend on your aims.
For us, our aim is to get the imagery onto Mapillary. Therefore, we assign leaders to each team. These leaders are responsible for making sure the imagery is uploaded.
Leaders (or those they’ve nominated) upload the imagery to their own Mapillary account using the Map the Paths Uploader. We invite these Mapillary accounts to join the Trek View Mapillary organisation in order to keep track of all the images centrally.
People are busy, so don’t expect them to get the images online overnight, but you should set a reasonable deadline for imagery upload that the team leader is responsible for meeting.
I find setting a group video call with attendees a week or two after the event with the aim of going through imagery sets a good deadline for people to work towards.
During such a call, go through some of the highlights for the day. Share stories. Take a look at some of the object data automatically abstracted by Mapillary and added to the OpenStreetMap database. Ask attendees to update missing OSM data for the areas covered during the image mapping party (e.g. tag all the surface types in the imagery). Look for areas that could be covered in a future mapping party…
Ready to run your own image mapping party now?
Let me know about it. I’d be more than happy to help promote it.