Yet another cheap thermal imager incoming.. Seek Thermal

[johntra]

Thanks Frenky, its good that someone knows they're way around this monster thread...

It seems that these are still work-in-progress and not really ready to be incorporated into a project, would you agree?

Looking around I've found a project done with the FLIR-one camera (lepton) connecting it to a ras-pi. This seems to be a mature project (albeit being a hack).
The resolution is lower, but it seems that the quality is higher?
http://www.pureengineering.com/projects/lepton

Ideally I would want a solution that works with the least amount of tinkering with the thermal hardware and code. (Leave that to the specialists and focusing on the project itself).
Does anyone have experience with incorporating any of these solution into a hobby project? What are the considerations?

[frenky]

If you need something more professional take a look at official SDK for lepton (bottom left on the site):
http://www.flir.com/cores/display/?id=62648

[johntra]

Thanks. It seems that SDK is woefully under-documented.

Am I correct in the conclusion that there is just no simple, integrable(OEM), thermal camera that's affordable (<1000$) and has an official SDK?
Either board-level or boxed.
Does that exist?

[frenky]

There are some options:

Therm-App seems to have a SDK:
https://therm-app.zendesk.com/hc/en-us/articles/202707182-Is-there-an-API-for-Therm-App-

and new Flir One will have it:

[frenky]

And if you need something that works out of the box, this new Flir AX8 might fit the bill:
http://www.flir.com/automation/display/?id=65816

[osip]

It is rather convenient though and runs well (as well as it can) with my $49.99US Digiland 7" tablet.

   ...ken...

Ken, Could you please specify which exact model Digiland 7" are you using?

I am looking for a cheap tablet to run Seek Thermal. I want to try using it for lab research (working with CO2 lasers).

I bought an iNova EX780, and the app wouldn't work. It just doesn't allow me to download me from Play, and if I side-load it, it starts and crashes.
The support team was not very helpful.

Any information on what tablets (preferably inexpensive) work well with Seek Thermal will be appreciated. Thanks!

[cynfab]

osip,

I have a Digiland DL701Q that I got at BestBuy around Xmas time. It was on sale on Black Fri for $39.99, but I missed that and paid $49.99.
I use an USBOTG extension that I got on e-bay. 5 wire not 4 wire. It was more expensive by about 2x, but still at ~$5.00 it was a bargain.

It has been running with each rev from Seek, and is now at 1.9.0/1.3.0.0/1.8.0.0 (App/CameraSW/ImageProcSW)

The key points for a useable tablet are the USB OTG, Camera, Google Play, Android 4.4 or better. The Seek App will NOT run properly without a camera that faces away from the screen. Some call this front facing, some call it rear facing.

   ...ken...

http://www.bestbuy.com/site/digiland-7-8gb-black/8610212.p?id=1219354106671

[arsenix]

Can anyone comment on the best embedded driver for the Seek right now?

I have been playing with libseekthermal https://github.com/ethz-asl/libseekthermal which was very easy to get working but does very basic dark image subtraction on the image. It will need quite a bit of work to get to usable images (dead pixel removal, hotspot filtering etc). It looks like most of the efforts from folks here are windows applications?

I have read this entire thread, which rivals some novels in length! Seems like there are still quite a few unknowns with regards to calibration and cleaning up the images. Has there been any word from Seek as to when their SDK is going to come out? Maybe the next edition of the Lepton will be a better prospect for embedded projects in the near term?



James

[miguelvp]

Can anyone comment on the best embedded driver for the Seek right now?

I have been playing with libseekthermal https://github.com/ethz-asl/libseekthermal which was very easy to get working but does very basic dark image subtraction on the image. It will need quite a bit of work to get to usable images (dead pixel removal, hotspot filtering etc). It looks like most of the efforts from folks here are windows applications?

I have read this entire thread, which rivals some novels in length! Seems like there are still quite a few unknowns with regards to calibration and cleaning up the images. Has there been any word from Seek as to when their SDK is going to come out? Maybe the next edition of the Lepton will be a better prospect for embedded projects in the near term?


James

There is also jadew implementation he released just yesterday. Haven't had a chance to play with it because I'm on the middle of another project.

https://www.eevblog.com/forum/chat/seekthermal-how-to-correct-the-image-received-from-the-sensor/msg696895/#msg696895

No news on Seek's SDK yet, I do check every now and then and I did send them e-mails and requests but nothing.

[arsenix]

Excellent! I just pulled that down and built it. Definitely looks better than the libseekthermal I was using!

After reading this thread I am unsure if the Seek Imager will really be suitable for what I want to do. I am trying to do aerial thermal imaging from a moving drone. I don't need very high resolution, but the frame averaging and noise from the Seek is going to make things challenging. Even though the Lepton is lower res it may be better. Any thoughts?

[miguelvp]

Excellent! I just pulled that down and built it. Definitely looks better than the libseekthermal I was using!

After reading this thread I am unsure if the Seek Imager will really be suitable for what I want to do. I am trying to do aerial thermal imaging from a moving drone. I don't need very high resolution, but the frame averaging and noise from the Seek is going to make things challenging. Even though the Lepton is lower res it may be better. Any thoughts?

If I have time I wanted to do a wavelet transform of the image, at the current frame rate it should give plenty of time, unlike an FFT that would take just too long, but right now it's on my low priority list.
Maybe someone else can give it a whirl

[Fraser]

@arsenix,

For moving platform image capture it is recommended to use a thermal camera with >25fps.
60fps is excellent in the application. The challenge is in finding such a camera that is small and light enough for the application, such as a FLIR Tau. The FLIR Lepton 80x60 resolution camera is both too low resolution and frame rate for most aerial applications. Decent optics are also needed and the Lepton is not suited to such a task. IMHO neither is the higher resolution SEEK camera. The new Lepton 160x120 resolution core may offer better performance but you really need 320x240 resolution with a decent lens attached.

Thermal cameras with decent resolution and frame rates higher than 9fps are controlled technology and this makes their procurement outside of the USA difficult, especially if you intend to fit it on an aerial vehicle ! The US DoD do not like the idea of high performance thermal camera equipped drones outside of their control 

You can use a 9fps camera on a drone and a search of You Tube looking for 'thermal camera drone' should get some hits. Take a look at the images and see whether the limited frame rate and resultant image blur is acceptable in your application. If so, you have a lot more choice when it comes to finding small, light cameras for the application. They will not be cheap though.

I would also recommend a quality camera stabilsation gimble in such an application. Quad and Octocopters are the most stable platform for using a low frame rate camera. A fixed wing vehicle or conventional helicopter are not great platforms as they either move too quickly across the target or they are too unstable in flight.

Aurora

[Fraser]

In case its of interest, I see that SPC have done a reverse engineering cost analysis on the SEEK camera and its 12um microbolometer.

http://www.slideshare.net/Yole_Developpement/yole-seek-thermalinfraredcameraandraytheonirmicrobolometersample

Sadly the figures are blurred but it is interesting to see that the seek product was worth someone's time to carry out such a RE task 


UPDATE:

The original PDF is here but still blurred content as they want over $3000 for the full report !

http://www.systemplus.fr/wp-content/uploads/2015/01/S-C_RS201_Seek-Thermal-Camera_Sample_V2.pdf

And the earlier version in case it is different:

http://www.systemplus.fr/wp-content/uploads/2015/01/S-C_RS201_Seek-Thermal-Camera_Sample.pdf

Aurora

[arsenix]

@arsenix,

For moving platform image capture it is recommended to use a thermal camera with >25fps.
60fps is excellent in the application. The challenge is in finding such a camera that is small and light enough for the application, such as a FLIR Tau. The FLIR Lepton 80x60 resolution camera is both too low resolution and frame rate for most aerial applications. Decent optics are also needed and the Lepton is not suited to such a task. IMHO neither is the higher resolution SEEK camera. The new Lepton 160x120 resolution core may offer better performance but you really need 320x240 resolution with a decent lens attached.

Thermal cameras with decent resolution and frame rates higher than 9fps are controlled technology and this makes their procurement outside of the USA difficult, especially if you intend to fit it on an aerial vehicle ! The US DoD do not like the idea of high performance thermal camera equipped drones outside of their control 

You can use a 9fps camera on a drone and a search of You Tube looking for 'thermal camera drone' should get some hits. Take a look at the images and see whether the limited frame rate and resultant image blur is acceptable in your application. If so, you have a lot more choice when it comes to finding small, light cameras for the application. They will not be cheap though.

I would also recommend a quality camera stabilsation gimble in such an application. Quad and Octocopters are the most stable platform for using a low frame rate camera. A fixed wing vehicle or conventional helicopter are not great platforms as they either move too quickly across the target or they are too unstable in flight.

Aurora

I am familiar with the Tau and some of the other "Full Motion" thermal imagers. I don't really need that type of resolution thermal imaging or frame rate. I need about half a degree pixel size at 30-45deg FOV, which is more like a high resolution thermometer. 1Hz is fine, but the motion blur may be challenging to deal with. The Lepton resolution is probably adequate, but more is always better. Looking at videos online it seems that the Lepton has much less motion blur than the Seek. Seek seems to be averaging a number of frames at a higher rate to reach 9FPS. I have even considered using the Malexis 16x4 thermopile array, but that is a bit too low resolution for what I need.

Appreciate the tips. I will play around with the Seek a bit more as I like the packaging but I may have to switch to the Lepton. Lepton is a bit more expensive and lower res but seems to be higher quality. The Lepton2 may be a good upgrade path as well...

[TinWhiskers]

The new Flir one for IOS is apparently available now at $249.  Range is -20 to 120° C.

[Fraser]

Comment on 2nd Gen FLIR ONE removed as OT

Aurora

[Trax]

Are there any data sheets for the new lepton sensor available?

Does anyone know if this price is with or without the tax?

[arsenix]

It is strange that they are not really publishing any low level specs on the sensor. At least no discussion of the resolution. Maybe they aren't sure yet if it is shipping with a new lepton sensor?

Very interested in this although given that I want to interface it with Linux and not Android it could be a science fair to do what I want. Might be a while before the new Lepton sensor is available as a standalone module.

[Trax]

I found a quotation on RS-Components the price there is € 229,00 + tax thats relay cheep

But I'm also worried about the resolution of the used lepton sensor if they would use a new one I guess they would advertise it every ware.
I was on Tuesday on the LASER World of PHOTONICS in Munich at the Flir stand, they did not have any new infos on the Flir one.

[Fraser]

Comment on 2nd Gen FLIR ONE removed as OT

Aurora

[miguelvp]

Avoiding specs not to compete with their higher range products?

Nah, they wouldn't do that.

I guess they don't want to attract the attention of people that use their other professional offerings.
Of course there is more to the stand alone ones than the tech specs, but why would someone spend $1000 when they can do the same with better resolution for just $250?

As for Seek, They need to drop $100 off each of their offerings, even if the new Flir One's software sucks on launch (not saying it does, only if it does)

[jadew]

I only recently got my Seek Thermal and I wasn't sure if it's the right choice, but the reason I went with it is the lack of info about the new Flir One. The info they provide sounds too much like marketing wank and too little as real resolution improvement over the old one. My guess is that it's the same sensor and the same thermal resolution and they only improved the video camera.

[Trax]

My guess is that it's the same sensor and the same thermal resolution and they only improved the video camera.

Exact my thoughts if they don't publish the exact 111x222 spects of the microbolomether a 4x can mean anything, and its possible that they just improved the normal camera it was VGA so now it would be 2560x1920 so just some sheep 4 MPixel sensor.

If the new lepton would really be 4x better resolved that would mean 320x240 and thats much more than some of their higher end products have

I only recently got my Seek Thermal

Are you happy with it?

[Fraser]

Comment on 2nd Gen FLIR ONE removed as OT

Aurora

[arsenix]

I just preordered one... we'll see what happens!