Closed dakotabenjamin closed 8 years ago
"Put GPS positions in EXIF based on GCP" If I correctly understand this, and code, right now it should be possible to create orthophoto from my dataset with GCP?
I've tried Boruszyn dataset with gcp_list.txt and my output is like this:
Script generated ../odm_data/boruszyn/reconstruction-with-image-size-2400-results/option-0000_georef.ply correctly, but for GeoTIFF and LAS there is no output.
quitting cause: "/home/mechanik/dev/OpenDroneMap/bin/odm_extract_utm" -imagesPath /home/mechanik/dev/odm_data/boruszyn/ -imageListFile /home/mechanik/dev/odm_data/boruszyn/reconstruction-with-image-size-2400/pmvs/list.rd.txt -outputCoordFile /home/mechanik/dev/odm_data/boruszyn/reconstruction-with-image-size-2400/odm_georeferencing/coordFile.txt returned with code 256.
What I'm doing wrong? ;)
Did you run with the tag --odm_georeferencing-useGcp True
?
I'm getting the same error too, also during keypoint matching.
Hmm.. I've rerun with this tag, and:
Restarted whole process, but:
Maybe problem lays in coordinates of GCP? They're in local polish geodetic system (EPSG:2180), not in geographic ?
imagesPath was set to: /vagrant_data/odm_data/boruszyn_kap/reconstruction-with-image-size-2400
imageListFile was set to: image_list_tmp.txt
outputCoordFile was set to: utm_coord_tmp.txt
Error in OdmExtractUtm:
No valid GPS position found
Above is my odm_extracting_utm_log.txt. This error is thrown on line 318 of odm_extract_utm/src/UtmExtractor.cpp
It looks like that function, UtmExtractor::parsePosition, only looks at the jhead output to parse position, and the boruszyn data has no ephemeris, so it fails there.
I think the thing to do here is to write a condition to only run the knn-preselect if gps data exists in the exif. For now, you should run with --matcher-preselect False
Ran command ../../run.py --odm_georeferencing-useGcp False --start-with odm_georeferencing received error.
This is on the benchmark sample set useing a fresh ec2 r3.8xlarge instance.
quitting cause: "/run/shm/OpenDroneMap/bin/txt2las" -i /run/shm/OpenDroneMap/odm_data/benchmark/reconstruction-with-image-size-2400-results/option-0000_georef.ply -o /run/shm/OpenDroneMap/odm_data/benchmark/reconstruction-with-image-size-2400-results/pointcloud_georef.laz -skip 30 -parse xyzRGBssss -set_scale 0.01 0.01 0.01 -set_offset 440138 4588397 0 -translate_xyz 440138 4588397 0 -epsg 32617 returned with code 256.
I really should remove the benchmark dataset -- it is has poor geometry and rarely resolves well. I'll be posting some new and decent datasets in the next few weeks. In the mean time, a better dataset is Seneca.
"Could not open lasreader" -- did the install successfully complete after you pulled down the latest changes?
It looks like the issue is due to local-gen on the ec2 instance, will install and try again. Error in Georef: locale::facet::_S_create_c_locale name not valid
This was with seneca dataset
Tryed again, this time setting export LC_ALL="en_US.UTF-8" and run the seneca dataset Finished with no errors but this is the odm_orthphoto.png
@dakotabenjamin had a similar result on an earlier run -- a subset of the total area. He didn't see it later when he re-ran, so something strange is going on. Any thoughts, Dakota?
I tryed a re-run but had the same result. Probably related to issue #115 In the pvsm folder the list.rd.txt only contains the 4 images, same as his issue
./IMG_0512.jpg ./IMG_0513.jpg ./IMG_0544.jpg ./IMG_0596.jpg
I have a hunch that the problem lies in the Bundle2PMVS and Undistort processes run after bundler, but I haven't been able to isolate further.
On Saturday, August 29, 2015, Tom Lincoln notifications@github.com wrote:
I tryed a re-run but had the same result. Probably related to issue #115 https://github.com/OpenDroneMap/OpenDroneMap/issues/115 In the pvsm folder the list.rd.txt only contains the 4 images, same as his issue
./IMG_0512.jpg ./IMG_0513.jpg ./IMG_0544.jpg ./IMG_0596.jpg
— Reply to this email directly or view it on GitHub https://github.com/OpenDroneMap/OpenDroneMap/issues/154#issuecomment-135988483 .
Dakota M. Benjamin Case Western Reserve University, 2015 dakota.benjamin@gmail.com dakota.benjamin@case.edu
"And now here is my secret, a very simple secret: It is only with the heart that one can see rightly; what is essential is invisible to the eye."
Should cc @petrasovaa, on this, as this branch has LAS output of georeferenced point cloud, a feature she requested.
I don't think this is a factor, but when I ran the Seneca set successfully, I had done it on a native ubuntu machine (not a VM), also using --resize-to 1200
@merkato and @darkpioneer could you open up new issues for these and tag them 'python'?
Hello, I install with python branch but install.sh is deprecated, this install ceres-solver and google-glog don't exist
me too @julianandres
Closing as the python-port has been merged. if you have any more issues, please open a new one.
Hey all,
There is a new branch called python-port. This is a big leap forward that should resolve a few issues, but it needs plenty of testing before we switch over completely.
These are the changes so far:
Known problems:
Big thanks to the Spotscale folks for providing the code!