Extend Carto importer to process stitching of images

splashblot / dronedb

Location Intelligence & Data Visualization tool

http://carto.com

BSD 3-Clause "New" or "Revised" License

0 stars 0 forks source link

Extend Carto importer to process stitching of images #3

Open jjmata opened 7 years ago

jjmata commented 7 years ago

The idea is to leverage what's already in Carto, and queue a job there that stitches assets and creates a visualization using that tileset as a custom basemap:

[ ] Use Carto's import to explode the .zip and decide what's inside is meant for ODM
[ ] Batch process the job (* kick off ODM job via ECS tasks?)
[ ] Automate the creation of a visualization that once the import is completed adds the custom basemap automagically (and maybe even a layer of discrete points of image GPS locations used in the stitching from EXIF coordinates ... see here)

Thoughts? Comments @apercas or @ivanprado?

jjmata commented 7 years ago

Not that I trust what is there, but some of what's described above (for the first checkbox) was attempted in the importer-ODM-zip branch.

jjmata commented 7 years ago

Also related, this is what WebODM currently does in terms of actual workflow:

1. python /code/run.py --use-opensfm-pointcloud --orthophoto-resolution 100 --project-path /var/www/data/8c2d875b-3069-4301-b26e-7b3ede3d1838 --pmvs-num-cores 4
2. gdal2tiles.py -z 16-21 -n -w none data/8c2d875b-3069-4301-b26e-7b3ede3d1838/odm_orthophoto/odm_orthophoto.tif data/8c2d875b-3069-4301-b26e-7b3ede3d1838/orthophoto_tiles
3. PotreeConverter data/8c2d875b-3069-4301-b26e-7b3ede3d1838/odm_georeferencing/odm_georeferenced_model.ply.las -o data/8c2d875b-3069-4301-b26e-7b3ede3d1838/potree_pointcloud
4. Imported orthophoto /webodm/app/media/project/1/task/1/assets/odm_orthophoto/odm_orthophoto_4326.tif for Task ID: 1

See WebODM issue 89 as well.

jjmata commented 7 years ago

FWIW, just added exif-to-geojson to our image. This should help turn around quick visualizations that include the location of images that have been stitched.

This might not be the winning strategy and maybe just using pyexiv2 is better than a full-blown Node utility. See what ODM does with it here.

jjmata commented 7 years ago

We should probably update WebODM as well, and start a monthly rebase of it as well as we do for Carto.

See here for their progress.

jjmata commented 7 years ago

Moving to next milestone.

jjmata commented 7 years ago

Not critical for April, so moving to next milestone.

jjmata commented 7 years ago

Taking out from milestone plan indefinitely.