Can't track 3 objects with Vega

[thompson318]

In GitLab by @MattClarkson on Jan 31, 2020, 22:53

Hi there Steve,

@JoaoDanielMarquesRamalhinho and I were struggling today. SmartLiver in python can track 2 objects reliably. But when we add a 3 object (NDI pointer), the code runs when the 3rd item is out of the volume, but crashes when the tracked pointer is first made visible. The program does appear to crash when decoding a response from the tracker (i.e. low level ndiCommand).

Are you able to look? Soon as poss?

Thanks

M

[thompson318]

In GitLab by @StephenThompson on Feb 2, 2020, 10:37

My first thought is conflict between two of the tools. Have you tried all three combinations of two tools? Have you tried tracking them using NDI's software? I think the old smartliver PC in in the OR has working software on it.

Does surgerynditracker crash or just fail to track?

[thompson318]

In GitLab by @MattClarkson on Feb 2, 2020, 18:03

all 3 track ok in NDI software, which is installed on cybernet.

surgerynditracker crashes.

[thompson318]

In GitLab by @StephenThompson on Feb 4, 2020, 15:02

I'm able to replicate this, using scikit-surgerytrackervisualisation. Error is:
Traceback (most recent call last):
File "C:\Users\SmartLiver\Desktop\scikit-surgerytrackervisualisation\sksurgerytrackervisualisation\overlay_app\overlay.py", line 78, in update
self._update_tracking()
File "C:\Users\SmartLiver\Desktop\scikit-surgerytrackervisualisation\sksurgerytrackervisualisation\overlay_app\overlay.py", line 97, in _update_tracking
port_handles, _, _, tracking, quality = self._tracker.get_frame()
File "C:\Users\SmartLiver\AppData\Roaming\Python\Python36\site-packages\sksurgerynditracker\nditracker.py", line 405, in get_frame
ndicapy.ndiCommand(self._device, self._capture_string)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x81 in position 2: invalid start byte

I don't know what's causing it yet.

[thompson318]

In GitLab by @MattClarkson on Feb 4, 2020, 15:05

Thats the fella.

[thompson318]

In GitLab by @StephenThompson on Feb 4, 2020, 15:39

mentioned in commit 8c383e0

[thompson318]

In GitLab by @StephenThompson on Feb 4, 2020, 15:44

I can fix this by setting the system to use "TX transforms". See commit above. I'll do this for now and push a new release. TX transforms use text messages between the tracker and the computer, as opposed to BX message s which are binary. So BX should be faster, but for the frame rates we use (< 30 fps) it really shouldn't matter. I'll raise another issue to try and investigate this further at a later date.

[thompson318]

In GitLab by @MattClarkson on Feb 4, 2020, 15:45

Great. Thanks.

[thompson318]

In GitLab by @StephenThompson on Feb 4, 2020, 16:05

I've just released scikit-surgerynditracker 0.1.7. Please can you test SmartLiver with it and if OK close this issue. I have raised issue #15 to do a fuller investigation when time permits.

[thompson318]

In GitLab by @MattClarkson on Feb 4, 2020, 16:05

@JoaoDanielMarquesRamalhinho - over to you.

[thompson318]

In GitLab by @JoaoDanielMarquesRamalhinho on Feb 4, 2020, 17:09

Just tested, it works.

[thompson318]

In GitLab by @StephenThompson on Feb 5, 2020, 12:15

closed