Trouble with correspondences

[Adetola Ololade Nicole Koiki]

Hi all,

I am a graduate student at the University of Maryland, College Park. I am using PyPTV on Windows (Anaconda environment) with a 2-camera setup and am encountering a crash during the correspondences step. The GUI launches correctly, and I have calibrated successfully. I can run Start → Init/Reload, Preprocess → High Pass Filter, and Preprocess → Image Coord without issue. Particle detection works in both cameras. However, when I select Preprocess → Correspondences, the GUI immediately closes. The command prompt prints:

short_file_bases: [WindowsPath('img/cam1'), WindowsPath('img/cam2')]
Warning: No sequence parameters found, skipping target writing
Frame 123456789 had [7] correspondences.

My images are named img/cam1.0303 through img/cam1.0914 and img/cam2.0303 through img/cam2.0914. In the GUI under Sequence processing, I have set First = 303, Last = 914, and the basenames to img/cam1 and img/cam2. The .ori files exist and remain unchanged between launches. I have tested multiple correspondence criteria (eps0 from 0.1 to 1.0, relaxed cn/corrmin values) and expanded the 3D volume bounds, but the number of correspondences remains around 7, and the GUI still crashes.

My main question is why the correspondence routine reports “No sequence parameters found” even though they are set in the GUI, and whether this crash could be related to the small number of correspondences in a 2-camera configuration. I am trying to determine whether this is a configuration issue, an orientation problem, or something else I haven't considered.  Any guidance would be greatly appreciated.

Thanks,

Adetola

[Alex Liberzon]

Dear Adetola

Please share a small subset of the working folder so we can test - it's difficult to say what's wrong. I guess the something changes from the 1st pair of images you do in preprocessing manually to the sequence. 

Alex

[Adetola Ololade Nicole Koiki]

Hi Alex, 

Here is the attached working folder of data. Thanks for you help. 

Best,

Adetola

[Alex Liberzon]

Alex Liberzon

9:12 PM (0 minutes ago)

to Adetola, openptv

Hi

There were some small mistakes in the parameters: a) instead of highpass, it was "using existing targets" checkbox, b) it's better to have glass = 1.0 mm rather than 0.0, c) in sequence, you need to provide the full path, i.e., img/cam1. and not cam1. d) In the new PyPTV, we use img/cam1.%04d instead of img/cam1.#, e) the volume is incorrect - your calibration body is 0 - +1500, but then the volume was -700 to 700, the rest are all kinds of small tweaks that I cannot really say are optimal, but working. So you'll need to keep adjusting the parameters to improve them. It's difficult to capture both slow and fast particles - there's some search space to explore for tracking. 

Nowadays, one can run PyPTV without installing it: ) so in your folder, you can just try:

uvx --python 3.11 pyptv .

(note the period or just add there the full path of your folder)

[Alex Liberzon]

https://limewire.com/d/jZ8BT#UkQKIABJsX

[Adetola Ololade Nicole Koiki]

Alex,

Thank you for the changes! Is there a reason the GUI crashes when I run correspondences? I ran my data with your changes and I get more correspondences, but the GUI will close with no error message in the command prompt.

Best,

Adetola