Using the Seek Thermal Compact Pro in an automated test setup

[nsd_c]

Hi,

I'd like to use the Seek Thermal Compact Pro in an automated test setup. Basically, I want to be able to capture individual frames from the Seek on a PC on command. I would like to be able to read particular pixels from those captures and record the temperatures of those pixels. What's the best way of going about this?

I see that there are a number of different libraries to choose from on Github--one user was nice enough to compile a list of those here, though it looks that a number of these applications are entirely unable to indicate what a given pixel's actual temperature is (?), which is useless for my application.

I have ordered a Seek Thermal camera but I have not yet received it, so I have not yet had a chance to play around with JoeC's software (or any other SW for that matter). I see that the seek thermal interface for his software is a separate download(?); that download contains some C Sharp (??) source files as well as a .DLL. Is it possible to link against this DLL to build an application to do what I want, or are there other SW components that I need that aren't contained in that DLL?

Any ideas/suggestions? What's the best library/SW package for my application of all those available?

[Spirit532]

As far as I know there's no way to get accurate temperature readings from any seek camera using custom drivers, other than manually performing an external calibration every time it starts, and then once in a while.
For automated testing you would've been much better off with a used FLIR Ax0 or Ax5 series camera(A40, A30, A35, etc) that were designed for this. They're a bit more expensive, but the noise can be a lot lower.
Unless you're only looking for large temperature jump pass-fail testing, you're unlikely to get anything seriously useful out of it.

JoeC's and other software contains all the code you need to initialize the camera and get the frames out of it, as well as detecting the internal shutter frames for NUC.

[nsd_c]

Thanks for the information, as well as for the FLIR model suggestions. Those might be within my price range on the used market.

As far as I know there's no way to get accurate temperature readings from any seek camera using custom drivers

Is this because nobody has reverse-engineered this part yet, or does the Seek simply not communicate this information over USB...? It would seem crazy to me that this isn't possible as the mobile app has to know temperature somehow, right...? If it is a matter of "just hasn't been reverse engineered yet," is this because the problem seems horrendously difficult, or is there just not sufficient interest? (I am wondering how I might fare trying to attempt reverse engineering this... I suspect that many others much smarter than myself have already tried and walked away.)

[...] other than manually performing an external calibration every time it starts, and then once in a while.

What is required to perform an external calibration? Do I need an IR thermometry cal standard? How often is "once in a while?" I have seen a thread on this forum about making DIY cal standards, but I don't think that I could justify buying a "real" one. Perhaps someone could point me to where I could read more about calibration the Seek to use with JoeC's software...?

Unless you're only looking for large temperature jump pass-fail testing, you're unlikely to get anything seriously useful out of it.

Do you mean with the Seek I should not expect better than pass-fail testing? Is this because of noise issues? My hope was to produce plots of temp vs. time at a marker.

[IwuzBornanerd]

The Seek cameras themselves do not provide any temperature values for the pixels; that has to be done in software.

I think most people lost interest in the Seek cameras due to the high noise of the early Seeks and the availability of other cameras that are lower noise than even the newer Seeks.  I have observed some disturbing anomalies in my Seeks that should concern you, such as on occasion the temperatures of the entire scene will suddenly jump a couple degrees Fahrenheit, stay there a short while & then drop back to where they were.  I have not tried to figure out what is going on there but it is either something in my code (but I don't think so) or maybe the thing for some reason shifts the bias on the entire array (since it happens after a shutter frame).  I am quite sure I have seen this at room temperature but it is more frequent and more pronounced when the cameras are out on the windowsill in freezing temperatures.

I have a method for getting temperatures out of Seek data but I don't consider my results "proven".  I get readings that I think are good enough for casual home use but I have not proven how accurate or consistent they are and I believe my process needs to be tweaked as it is.  If you expect your numbers to be accepted by customers or even your own employer I think you would want to prove the accuracy and consistency of the results via some recognized "standard" equipment.  You should also note that even the FLIR A40, 50, 60 series mentioned above is specified accurate to only +/-2*C (or 2% of reading), so you can't expect any better accuracy than that.

My method is based on 2 things from the camera:

a) There is a pixel in each frame that provides a measure of the temperature of the camera (pixel 1 in the non-pro).  The number is different in each camera but it decreases by 10 for every degree Fahrenheit increase in camera temperature.
b) One of the frames put out on startup, the one with status byte=9, contains a curve that looks much like a radiance vs. temperature curve.

If you would like details on how I derive temperature from Seek data I prefer that you post that question on my thread:
  https://www.eevblog.com/forum/thermal-imaging/running-seek-thermal-cameras-(-others)-on-a-raspberry-pi/?PHPSESSID=ti38hc7ih08p4l3svtjo319sm2 
as that is my forum for talking about my software and I think people would be more likely to look for & find the information there.  OR maybe you could start another topic about how to determine pixel temperatures on Seek (all?) cameras.