Agisoft Metashape
Agisoft Metashape => Bug Reports => Topic started by: macrog on February 02, 2015, 07:19:56 PM
-
Hello Kjellis85,
Do you mean coded targets that were printed from older versions? If you do not wish to re-print new targets from latest 1.1.0 version, then you should check on "Disable parity" option in the Detect Markers dialog.
Hi Alexey,
Please check if this option works, because i've noticed that in my case it doesn't work at all - I have set of photos with "old version" markers and I could not detect markers with this option checked on
regards,
Mac
-
Hello Mac,
There's additional check in PhotoScan 1.1: at least two black sectors should belong to different semi-circles, so if you have all the black sectors situated in the same half-circle, marker detection will fail for old version targets.
-
Hi,
ok - i understand - so in my case it wouldn't work - i have there markers no. 1,3,5,7 ;) thanks for explanation
-
Hello,
I confirm there is a bug with the detection of old coded targets! I tried to detect old targets with photoscan 1.1.3 by checking and not checking "Disable parity". For example, the old targets 71,603,255,437 and 85 are detected with PS 1.1.3 (Disable parity checked and not checked) but the numbers of the targets are wrong (respectively: 1,97,45,78 and 5). On the other hand old targets such as 361, 35, 21, 143, 939, 5 69 and 2015 are not detected by checking or not checking"Disable parity".
Please could it be possible to detect old markers in the new release? I have old targets printed in vinyl that I would like to reuse.
Regards,
Sebastien
-
Hello Sebastien,
Could you please send us this image for tests?
However, I can say that old targets like 5, 63, 21 would not be detected by the new version, as all the black sectors are situated in the same half-circle.
As for the target indices, new version target numbers corresponds to the new PDF file.
-
Hello
It would be very useful that we can keep as an option to have old target detection.
In my case data acquisition is costly and labor intensive, and it is important that I can re-use photographs taken previously, they represent information of high value. At the moment re-using them is much complicated because of target detection issues (one data set represents typically 200 photographs with 50 markers on each of them...)
Bruno
-
In the version 1.1 it should be possible to detect targets from the old version (with disabled parity option), unless all the black sectors are inside the same semi-circle.
Unfortunately, we were unable to reproduce the problem with "good" targets (generated in 1.0) detection in the version 1.1, mentioned by Sebastien.
-
Next 1.1 version update will fix this issue and targets from older PDF would be properly detected, except few first targets that have all the black segments in the same half of the circle.
-
Hello,
Version 1.1.4 has been released, so please check the datasets with the old targets (with disable parity option).
-
Hello,
I have checked the detection of old targets with the release 1.1.4. It is still giving wrong numbers for the targets!
Regards,
Sebastien
-
Hello Sebastien,
Old labels cannot be extracted, since the indexing starts from new version targets (with parity) anyway. And when you disable the parity the list of targets without parity is added to the end.
But have you checked if the targets are now recognized?
-
I too am having a similar issue.
I am running PhotoScan 1.1.6 build 2038, and am trying to reprocess photosets shot using GCT's printed from a machine running Version 1.0.4 build 1847. The updated version is not detecting most targets and when it successfully detects targets, it detects them as having different ID numbers. I have disabled parity and the issue persists. Also, the targets I am using are in the 500-700 range of the old series. Is there a stable fix for this?
If not, is it possible to remap target IDs from the old series so that I can still use the older text files I have to import GCT coordinates? Or is it possible to revert to an older version of PhotoScan so I can process legacy data?
-
I also have this issue, 4 markers are not detected in PhotoScan 1.1.6 build 2038, and they all fulfil the criteria you mentionend (all black parts not being in the same semi-circle).
Also, why is it not possible to use the disable parity option when detecting markers using the Python API?
-
Hello beerockxs2,
Could you please send a couple of sample images with these markers to support@agisoft.com?
-
Hello beerockxs2,
Thanks for the images, looks like the targets are just too large compared to the image resolution. I've downscaled them, keeping the same resolution of the original image - they became detectable in 1.1.
At least in the version 1.2 there's NoParity flag for the detectMarkers() function.