Thanks for your feedback, oba510!

oba510 – Looks like we commented at nearly the same time!

I wasn’t planning to write out foot=no, but instead just remove the tag entirely – at least where sidewalks are mapped separately, since we don’t want routers directing people to jaywalk in the car lanes. But if you think foot=yes should remain on roads where the sidewalk isn’t mapped separately, I’ll leave it be.

Well, apparently it is not permitted by California jaywalking laws to walk around in the street if a sidewalk is available (and if you do that on Van Ness, you will definitely cause a wreck), so I’m gonna agree with Lejun that these foot=yes tags are probably old, and wrong, and should be removed.

“Pedestrians should use sidewalks and designated overpassed, tunnels and bridges whenever available in California. The only time it is legal for a pedestrian to walk in the road is if a sidewalk or passageway is not available and if the pedestrian walks on the leftmost edge of the road.” – https://www.damfirm.com/california-jaywalking-law/

But I’ll pop into the chats that SomeoneElse pointed out to avoid stepping on any toes.

Shoot, messed up the link. I got a bit reckless, since I forgot you can’t edit comments.

For some reason.

My only real problem with the Android Mapillary app – but it is a big one for me – is that I cannot opt to use my phone’s wide angle lens camera. I explain my feelings more, in a https://www.openstreetmap.org/user/JesseAKARaccoon/diary/395755:

“In terms of editing OSM, images taken with the standard lens on the Mapillary Android app can be nearly useless sometimes. I’ve had plenty of instances where I’ve been unable to verify street names, other signage, etc. Gathering house addresses, names of businesses, and other features not DIRECTLY near the road is often impossible. By the time something is close enough to see well, it’s out of frame or blocked by the vehicle ahead of me.

Plus, aesthetically, having imagery that only looks straight down roads is super-unappealing. It squanders Mapillary’s ability to stitch nearby photos together,” (not to mention, the user’s ability to see things other than the pavement ahead of you).

Honestly, I’m eager to try using a 360° camera to capture images, for basically the same reasons. And now I can, since I’ve found out how to use mapillary_tools! Just need to get a camera, first…

Though, before I started using mapillary_tools, I DID used to have a new folder for each sequence, and upload each folder one-at-a-time using the web uploader. They’d be called Trip1, Trip2, and so on. Thus, why my current folder is called Trip1, haha!

Oh, no, I put as much into that folder as I feel like! Notice the parameter in my command, “–cutoff_time 30.” That means that I simply stop capturing images for 30 seconds, the next time I start, a brand new sequence begins!

So I can fill the folder Trip1 with as much forward-facing imagery as I want, and process + upload it all when it’s best for me!

I will create additional folders for situations such as images taken with a 90° offset. Other than that, I just use the one folder, and clear it out when I’m done. It’s great!

To be clear, I don’t use the Mapillary app.

I capture images with OpenCamera, selecting my wide angle lens. I set the camera to repeat at the desired interval (usually 1 image / s). In settings, I make sure images are saved with GPS info. All normal images are saved to storage/dcim/Trip1.

If there are images I wish to treat differently (e.g. images taken with a 90° offset), I must save them to a different folder and process them with slightly different commands.

Then, I use the command line program Termux to run a command like

mapillary_tools process –import_path storage/dcim/Trip1 –cutoff_time 30 –duplicate_distance 1.5 –duplicate_angle 360 –interpolate_directions –offset_angle 0 –overwrite_EXIF_direction_tag –user_name jesseakaraccoon –advanced

And that’s about where my work ends!

I chose the parameters such as duplicate_distance and such based on experience, to minimize any further work. When taking photos by foot, I try to walk in the direction I shoot, so that the “–interpolate_directions” command provides accurate compass settings for me.

So… my workflow is basically “take pictures and enter one command.” Seems easy. But I think I earned it after all the difficulties and failures I had to go through to develop this process, haha!

I don’t have a super-strong opinion, here. I use Bing imagery, myself. Here in Humboldt County, California, at least, it usually lines up well with my GPS traces and such.

Sure! I have new entries with how I installed mapillary_tools, and how well it’s working. It’s reduced my effort with processing images significantly.

NEVER MIND, I JUST INSTALLED mapillary_tools! And I’m excited to give it a try!

Okay, so…

I found out how to run command lines and install basic Linux packages on my phone without rooting it.

BUT

You need Python 2 to install mapillary tools (3 won’t work), but Python 2 has been so thoroughly retired in favor of Python 3, that I can’t use it nor mapillary_tools without rooting my phone…

Although I love Mapillary as a concept, and will continue to contribute, it feels like development of the Android app and mapillary_tools has been essentially abandoned.

@vorpalblade – Oh, I’d heard of that. Good thinking, I’d love to use it! Except… is running that on an Android device a straightforward process? I see installation info for Windows, iOS, and Linux. Oh, gosh, I’d have to root one of my devices, wouldn’t I…?

I do turn on the compass on OpenCamera, but it’s very often off by about 20+ degrees to the right, and sometimes off by much more in a random direction. Can’t seem to get better accuracy than that…

@GOwin – Sure thing! Here’s one of driving on a sunny day, on a road I liked (mostly). Images were taken every 0.5 seconds. I’ve since changed to every 1s because I’ve found that my GPS EXIF data updates slightly less often than twice a second, resulting in about 15% overlapping images. The overlapping took quite a while to clean up using my phone… :

Link – https://www.mapillary.com/app/?lat=40.863244080555575&lng=-124.07715188333333&z=17&tab=uploads&pKey=xK6aptjrElhWX52aIUxOHW&x=0.5193214895634117&y=0.4894374844345887&zoom=0&focus=photo

Image key – xK6aptjrElhWX52aIUxOHW

This next sequence was taken out on a hike shortly afterwards. OpenCamera set to repeat every 3 seconds. Just held my phone by hand,, walking at moderate pace:

Link – https://www.mapillary.com/app/?menu=false&lat=40.58335066666666&lng=-124.15694425000004&z=17&tab=uploads&pKey=BzBx5v51dnjUQOMN2syAvQ

Image key – BzBx5v51dnjUQOMN2syAvQ

That can be a very good idea for helping pedestrians navigate!

I took a quick peek at some of your work. Good job!

With sidewalks, it may be a good idea to look at the “sidewalk”=* key of the nearby streets. If all of a street’s sidewalks are mapped, adding the tag “sidewalk=separate” can help prevent someone from double-mapping the sidewalks.

You can also use “sidewalk=both/right/left” on streets that have sidewalks that you don’t feel like drawing – but you seem to be enjoying the task.

In any event, I hope you have fun contributing!

This was a very interesting read, thanks!

This definitely affirms things I’ve noticed, and my conclusion that OSM’s strength is keeping up with recent changes.

I’ve submitted multiple updates to Google Maps regarding one-way streets, non-existent streets, important driveways and pedestrian paths, etc which still have not been approved. But since I’ve been contributing to OSM, the area around me has become much better mapped.

OSM and Mapillary are projects I’m proud to contribute to!

Thanks! Yeah, I should / will.

vorpalblade – Thanks! Yeah, I know that stuff. I already have the camera set to forward, not compass – so you’d think that each image would point to the next image in the series by default. But nope, some photos still point off in a random directions. I suspect it’s because some perceived mismatch in the “computer vision” mapillary uses the create a 3D model of everything.

After getting a bit more experience, I think that with longer and straighter trips, the proportion of weird-direction pictures can go down to 1-ish percent, since the computer vision has more to work with.

Strubbl – Thank you, I will! >95% of the buildings there now were drawn by me, but it looks like I still have work to do. XD