Simple way to force geotagged still camera photos onto server


#1

Recently retired, my pastimes include walking (those 25km walks where you follow the arrows) and cycling (to walks, and other places of interest). Also use an Open Street Map based map in the Garmin Oregon 600 GPS, noticed quite a few discrepancies, decided to jion OSM to either place notes or amend. It would help to upload photos taken around the discerpancy, Mapillary was suggested as the right place to do so.
spent this morning and half the afternoon uploading some 400 pics to Mapillary, found Firefox crashed couple of times, and more than once only 20 of 33 or 132 of 212 pics were uploaded.
Using a five year old Intel i5-2500 with built in Intel HD 3000 graphics, Windows 10, 8GB memory in a notebook. And Garmin Basecamp to geotag the photos from a simple Coolpix S3600 camera. So far so good.
Have some problems with Mapillary, hope someone can help me here:
1/ uploads of be it 30 or 210 photos tend to stall, will not resume - not even after half an hour. One way to find out which photos were omitted is to ‘walk’ along the line of uploaded photos. move each uploaded photo to a new ‘uploaded’ folder, upload the rest again. Is there a way to let the Mapillary server sort which pics were already uploaded, which were not, please? And it would help if the server fetched the pics in date time order - that’s how they’re named, and it would then be easier to find which ones were left behind.
2/ After drag 'n drop those pics to the upload screen one is given the option to drag photos to the correct location : could I please also at this stage set the correct orientation, as it will save walking along the line of uploads twice.
3/ One needs to repeat the placing of photos if they were for some reason not uploaded first time around : could the server perhaps make and keep a note where I moved the photo, thus not waste my effort?
4/ Having placed the photos, but not yet oriented them, I get the option to “publish” : would prefer this to be “process”, as I will later need to point each individual photo in the correct direction. Only once that is done may my photos be published : this avoids a complaint from people who look in on my edits on OSM, in that the locations there are not the adjusted ones, and the angle is always due north, even after I have changed that.
5/ Visual effects : beautiful, to see how smooth the line between points adjusts itself once you have moved a photo to the correct spot, but also time consuming : I’d rather do without the smoothly moving line, see the dot in one place now, blink, and its in the new, correct spot.

Perhaps I need a new computer : what would be reasonable specs, please? Not just for now, but for some years into the future as well. Should an Apple do better than Windows? Would Linux outpace them? Which distro? Then for the hardware : a quad-core, or an octa-core? 16GB ram? would of course need a separate video card : would a Geforce 1060 powerful enough?

Some questions, and looking back over the past six weeks Mapillary will soon be rid of me, as the upload and point in the right direction process takes more time than cycling say 40km to where the walk is organised, walking 25km, and cycling back. In other words, Mapillary is a huge drain on my time. That’s why the title of this post is as it is : Just force the photos down the line, not dependent on the server dropping something, and going over the photos twice to place and orient them.

With best regards / met vriendelijke groet,


#2

Welcome to Mapillary. I am glad you are asking questions and keep doing that.

The easiest way to contribute to Mapillary is to use a smartphone (preferably Android or iPhone) and take the pictures using the Mapillary app. It will both add location and direction when you take the picture, and even upload when you get home to your wifi.

I have not used the web uploader myself and cannot provide help for it. I am a programmer and use the Python scripts. I have written about that in http://forum.mapillary.com/t/running-the-python-scripts/130 . The Python scripts will run in a very stable way on even very slow computers but you must use the command line to run themn.

The web client of Mapillary is very resource heavy and I have even had problems on very fast computers. I have not used a Mac, but a computer with Linux usually runs signicantly faster than with Windows, but only very few computers are sold with Linux, so you will have to install it yourself. If you are buying a new computer I would say minimum 16 GB of RAM, preferably 32 GB if it is not too expensive; and more is always better.
I don’t think the web client will run much faster if there are more CPU cors. It is probably the speed of each core that is more significant. You don’t need a gaming graphics gard, but the Mapillary uses the graphics card, so it has to be decent and not just a cheap built in one. But I don’t know a lot about graphics cards.

If you are suggestions or bug reports, the place to tell Mapillary about them, is at https://github.com/mapillary/mapillary_issues/issues . You have to create a Github account and you will have to submit a new issue for each suggestion or bug report. Then the developers will look at each issue. They does not read all posts in this forum.

I hope this answers some of your questions - have others more experience using the web client than me?


#3

Dag koninklijke

When you are using the web uploader and it stalls you have probably lost these uploads (because you can not ‘publish’ them with the final button). If you are lucky and restart the browser it is sometimes possible to publish or the add more images. tip: If not, you can go to legacy.mapillary.com/map/upload/im and publish already uploaded images using the button there.

I never use the webuploader when I upload more than 100 images because I also hate when it crashes [also have a cheap laptop dual core with only 4Gig memory]; you better try to learn the python tools. Or use a mobile phone with the mapillary app and its integrated uploader, never problems with that.


#4

Not possible yet. I also would like the possibility to set the correct compass orientation in the review stage of the uploader. Or at least the ‘normalize’ option, so all compass directions are pointing to the next in line. Now I have to do it after the upload with the web edit tool


#5

You don’t need super computers to contribute to mapillary, your current notebook specs are absolutely fine. More important is a good and fast internet connection when using the uploader. Try an internet speed test online and see what your upload speed is. I have a very cheap internet connection with 2 mbps upload (and 20 mbps download speed) and think 5 mbps upload speed is about the minimum you should have to work comfortable…


#6

From Blender I remember that Nvidia is the best for graphic cards.


#7

Certainly do not buy an Apple. That would cut you off from the world of the open, free and advanced.


#8

@koninklijke Thanks for sharing! I think some of this you and I discussed in other communications, but just briefly addressing it here as well so that other people can also see it.

I have taken note of your suggestions for the uploader and integrating more “correction” options to it pre-publishing. We do have the intention to improve the performance of the web uploader, too. Currently, it works best with Chrome, so working with other browsers is something that we want to pay more attention to.

When upload in the web uploader stalls and you refresh or exit, you should still see the part of the upload that was successful. You can then add the whole bunch again, because the uploader is capable of figuring out which photos have already been uploaded (you need to go to the “review” phase for that to happen). But remember that the “memory” is lost once you click to publish the photos. So it only works on the “bucket” of photos you gather up before pushing them to be published.


#9

Thank you for bringing the legacy site to my (and other users’) attention.


#10

Current -new- desktop has an AMD Ryzen 5 1600 hexacore cpu, and a Radeon RX 560D graphics card : noticed that the trail of pictures is displayed much faster than on the elderly dual core 2nnn series with HD3000 graphics; looking at utilities like Speccy or Process Explorer can confirm that the cpu is only used for a small percentage, while the more powerful graphics really does speed things up on the display front.