I uploaded maybe 15GBytes of mp4 movies (say 40K odd images) and 700MBytes (2K odd) of static images 36 hours ago via the CLI. They seemed to be a little slow on ingesting but are now in map update (as far as I want to scroll back) Still using version 10,something of the tools.
I don’t watch this closely though having used a duplicate and filtered source for my concurrent OSM work over years now.
I think we are back at one week delay.
I now started mapping before the pictures are in. I don’t want people to die because Mapillary is lagging behind.
Adding to this, the silly Android app and I cannot recommend Mapillary anymore.
I even gave it only one star in the Google review.
Hi @filipc - we are still recovering from a processing delay this week, but in parallel we’re also working on a revamped pipeline which will minimize delays in 2024. It’s a big effort, so thank you for your patience.
In parallel, we are continuing to work hard on the Android app, we’ll be introducing wide angle capture, more stats on the camera screen, and an updated user experience. We appreciate the feedback and hope to earn a good review from you in the future
Thanks for the flag - there are still hiccups in the system that in some cases lead to longer than expected processing times. We are working on this - hopefully your sequences will be processed and available shortly (if they are not already)
@boris For some time now, I am a bit confused about the disparity between the sequence status indicator and what data is actually visible on the map. For example, on some of my sequences the status indicator basically says complete (number of images in sequence on grey background) but the sequence’s point cloud data, traffic signs, and point features (which the latter both depend on point cloud data) lag behind, sometimes for weeks. Furthermore, there seems to be no observable pattern in the order in which sequences get point cloud data. So, I am confused as to what complete actually means. Is this intended?
Hi @GITNE , it is not intended but we are aware of the problem. We are working on a more robust processing flow that will improve things in the medium/long term. For the short term, in the next weeks we will trigger reprocessing of inconsistent clusters and that should fix the cases you pointed out.
We were seeing a delay with the “Map Update” step since October 31st. The good news is that we’ve solved the issue and the map should be updated within the next ~4 days. (Right now the uploaded images are visible in the feeds, but the map tiles have not been updated to include them).
Thanks for reporting and we are looking forward to having the map updating regularly again.