Good afternoon folks, I’m a new WebODM user, installed it earlier this week, did some tests with small datasets (40-80 images in each), moved up to 200-300 images and pretty pleased with the results (coming from a Pix4d background). I then thought I’d put WebODM through it’s paces with a 1541 image dataset, it’s not a good dataset as it was taken from video, so ropey overlap etc, but wanted to compare to Pix4d…
However…19.5hrs later and it’s still running…latest message (over 19 hours ago) was “WARNING:opensfm.pairs_selection:Not all images have GPS info. Disabling matching_gps_neighbors.” - which I got on the other datasets too.
Decent spec machine, and I’m happy to let it run if that’s all that’s needed, but simple question (with perhaps a not-so-simple answer) is…do we think it will complete, or has it given up and just not told me? Task Manager is showing as still using 95% of RAM, but little CPU/little to no GPU (which I don’t think WebODM uses anyway does it?)
Any thoughts gratefully received…and I know it’s a “how long is a piece of string” type question
Hi Ade, I’m a week old her too and have been doing the same tests with the data-sets here. I parsed 542 photos in about an 1hr.20m. What are your specs? I had to allocate more resources in Docker to speed things up a bit.
There will be so many people better positioned to answer this but I’d say its just a memory thing - we’re on the same CPU, I’m on 128GB RAM.
In Docker home page open the ‘cog’ settings icon and then choose Resources, it’s just a slide bar for men allocation, see if that makes a difference?
Have you tried the DATASETS section of the forum, that’s where I’ve been getting my practice files from.
Also, if you’re looking for more feedback more often, I really enjoy using the verbose and debug options/flags to get more messages (more frequently), which better helps me gauge progress.
Also, Sefton gave you a great overview of where to poke. I wouldn’t leave Windows with less than 4GB, though it’ll sit happily with that little free provided you’re not using Chrome/etc and just letting it sit and process.
Thanks both, will have a play with reducing the number of images etc - need to find that sweet spot of “how many can I comfortably process on this machine?” really so I can bear that in mind for future projects. THanks!
Hiya - also, I’d allocated 90GB not the full 128, I think its just a matter of balance. There is also an option to process it via a RAM site or something.
Saijin also helped me with a similar question regarding the release of the resources I’d allocated after I close Docker, it’s possible that once finished processing the images then you should reset Docker resources to minimum to release them back to your PC.
He’s got the special case of Docker on Windows (likely with WSL2 as the default backend), so it should release nearly all of the resources once the workflow is finished. Currently there is a bug where filesystem cached data don’t get released, but I hope that will be resolved shortly.