Need different diameter in tp.locate on different computer?

[hmx0979]

Hello everyone,

Recently, I'm trying to run trackpy on two of my computers.

The code runs good, but I need to use different diameters in tp.locate to find particles.

In my image, there are about 20 particles.

1, on my macbook pro, 2017, 13-inch

diameter = 5, gives the reasonable tracking result, and it runs fast.

2, on my mac mini (2020) with 24 inch monitor, better CPU, newer version.

diameter = 151, gives the same tracking result but it runs slow, if I use diameter = 5, I will get ~20K particles.

I wonder what causes the problem?

[nkeim]

Interesting! What other arguments are you passing to locate? And how are you reading the image files?

Nathan

On Mar 26, 2021, at 9:58 AM, hmx0979 @.***> wrote:



Hello everyone,

Recently, I'm trying to run trackpy on two of my computers.

The code runs good, but I need to use different diameters in tp.locate to find particles.

In my image, there are about 20 particles.

1, on my macbook pro, 2017

diameter = 5, gives the reasonable tracking result, and it runs fast.

2, on my mac mini (2020) with 24 inch monitor, better CPU, newer version.

diameter = 151, gives the same tracking result but it runs slow, if I use diameter = 5, I will get ~20K particles.

I wonder what causes the problem?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/soft-matter/trackpy/issues/653, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AAWY23PVHJPOF6TIN7VKGHTTFSHGPANCNFSM4Z3OHG5Q.

[hmx0979]

frames = pims.open('*.tif') f = tp.locate(frames[0], 5, invert=False)

That's my code reading tif files (converted from avi) and my tp.locate function

[nkeim]

Thanks! One thing to check would be the gray values in your images (for instance, peek at frames[0][0, 0]). Depending on what optional dependencies are installed for PIMS, sometimes PIMS returns integer gray values in the range [0, 255], and sometimes it returns floats in [0, 1].

I'm assuming that you have the same trackpy.__version__ on both computers!

[hmx0979]

Thanks!

You raised a good point, my computers don't have the same version of trackpy.

My MacBook Pro is 0.3.3 and my mac mini is 0.4.2.

I have upgraded trackpy on my macbook to 0.4.2, so they have the same diameter now. d = 151,

This solves the problem that different computer requires different number. However, the diameter in pixel (d = 151) is too big. My particles are be no bigger than a few pixels, but here if I use d = 151, it is going to be very slow.

My images are dark-field images, particles are bright, background black. The particles are only several pixels in diameter.

[nkeim]

The diameter should be close to the actual diameter of the particles. At that large diameter the coordinates are almost certainly meaningless.

Your particles should be circular blobs on a dark background. Is that true? It might help to post a sample here.

On Mar 26, 2021, at 12:53 PM, hmx0979 @.***> wrote:



Thanks!

You raised a good point, my computers don't have the same version of trackpy.

My MacBook Pro is 0.3.3 and my mac mini is 0.4.2.

I have upgraded trackpy on my macbook to 0.4.2, so they have the same diameter now. d = 151,

This solves the problem that different computer requires different number. However, the diameter in pixel (d = 151) is too big. My particles are be no bigger than a few pixels, but here if I use d = 151, it is going to be very slow.

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/soft-matter/trackpy/issues/653#issuecomment-808368892, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AAWY23LVVCA6GEPRT6FXCW3TFS3WXANCNFSM4Z3OHG5Q.

[hmx0979]

Hi,

Although I don't quite understand what causes the difference in diameter arguments, I think I solved the problem by adding minmass argument in tp.locate, this works fine for me, and it is fast.

Here is a sample image, it is polydisperse nanoparticles.

Well, I don't quite sure why diameter = 5 gives me about 20 particles in 0.3.3 but ~20K here, but adding a minmass argument filters out most of them.

Thank you very much

[nkeim]

In v0.4 we switched to minmass=0 as the default because there are other, more scale-free ways to filter. But it is perfectly fine to use it.

Be sure to check that the coordinates you do get are valid, as shown by e.g. tp.annotate().