FLIR One Gen 2 Android app that works with Intel Atom based platforms.

[Fraser]

As some owners of the FLIR One Gen2 Android have discovered, the official FLIR app will not run on an Intel Atom based Android platform.

FLIR have stated that they do not intend to support the Intel Atom processor, now, or in the future.

I own some HUDL tablets that use the Atom processor and was disappointed to hear this news as the tablets are quite powerful and have ad event display.
FLIR apparently wrote their app in a code that will only work with ARM based platforms.

Enter the new 3rd party app that has just been released.......

A new Android App for the F1G2 has been written and released by Georg who is a member of this forum. He has written the app in Java and not only does it run at a higher frame rate than the FLIR app on my Motorola Moto phones, but it also runs fine on my Atom based HUDL tablets

The App also provides the full 160x120 resolution of the F1G2 rather than the cropped version produced by the official App. I was also pleased to see the MSX overlay gone as it interfered with close up images that need an auxiliary lens.

Me thinks FLIR coders could learn a thing or two from Georg. He has produced his first release of the App and for me it is already better than the official FLIR offering. Impressive.

There is a thread dedicated to the app here

https://www.eevblog.com/forum/thermal-imaging/flir-one-android-app-with-full-lepton-sensor-resolution/

Georg released the App as soon as it was stable and usable. He intends to continue its development. I do not know about other members, but I would happily pay for such an improved App.

My sincere thanks to Georg for sharing his work with us.

Fraser

[cynfab]

Fraser,
 When I first got my F1G2 I posted:

I would bet that the issue is related to having the processing done by java code vs. Android Native Libs . Java is certainly easier and probably the first implementation. Followed by a recoding in C/C++ in a Native lib. I believe that Seek transitioned from java to Native Lib several rev's of their app ago.
There is also an issue with Native Lib support in Android Studio at the present time.
 
Sooo, maybe there is hope for the Flir One app to improve in the near future, once there is a groundswell of complaints bi the early adopters. (hint... hint... hint...)

   ...ken...

I noticed some performance issues on my cheap tablet, I attributed that to their not using native methods/libraries.
After looking at the Android SDK and building an app of my own, I realized that this was not the case.
Flir seems to use native methods/libraries to do their live palette view where they display 15 or so copies of the image each with a different palette.
Perhaps as well as doing some of the MSK work.
IMHO this is a total waste of CPU cycles, and causes serious performance problems.

Georg didn't bother with such foolishness (nor did I in the app I wrote but didn't distribute to anyone but tomas123).

The result far exceeds the original Flir One app.

Native methods/libs only run on the platform that they were compiled for (ARM/X86). I believe (not sure) that one can do multi platform Native methods/libs but I've never tried to do one.

Thanks to Georg again.
While pure Java isn't the best performer, it beats a poorly coded native method/library hands down in terms of portability.

   ...ken...

And while I'm at it Thanks to Flir for providing the SDK. ;>))

[gansgar]

It works. At least on Android. Because Android Systems include an emulator for other architectures. So can an Intel x86 Tablet use the Flir One App too, as long as every assembly code inside the native lib is recognised as a command. Fraser had luck. It works on his Android x86 based tablet, but I had bug reports, where the tablets CPU couldn't run the Flir SDK. Because of probably poorly programmed code.

To the performance issues: I first programmed the current version completely in Java and used the Flir SDK only to get the Raw values. But I will switch in the next version to OpenCV (clearly including all processor architectures). That will speed up the code by a lot, because I will render everything with multithread CPU and GPU support.