Therm-App Info

The device I'm using is a Therm-App: http://therm-app.com/

The somewhat confusingly-named Therm-App is not an "app", but a hardware device that communicates with an Android device via USB. To use the Therm-App, you install a software app from Google Play store, called "Therm-App: Basic". The makers of Therm-App have made available their SDK to encourage development of other apps for the device, and a few third-party apps are starting to be available.
The Therm-App with a 19mm lens, attached to the back of a Nexus 7 tablet.
By default, the device comes with a 19mm lens, but a number of lenses are now available with the following specs:
Focal LengthRelative ApertureField of View35mm equivalent
35mmf/1.1410.5° x 7.9°200mm
19mmf/1.119° x 14°105mm
12.8mmf/1.028.8° x 21.7°70mm
6.8mmf/1.4155° x 41°35mm
The standard 19mm lens has pretty good range for detection, though not detail unless the subject is pretty big or pretty close. The 35mm lens is expensive, but after finally deciding to splurge, I've had no regrets: the increased range has yielded better detection and details. The trade-offs are a narrower field of view and more dizzying "camera shake".

In The Field

Using an infrared camera in the field is akin to using a metal detector: after the novelty factor wears off, it quickly becomes tedious with only the prospect of discovery motivating further usage. What happens is you end up staring at the screen the whole time, sometimes having to stretching yourself to stare at high angles:
To allay this, you can use a longer cable to point the camera independently of the screen:
a situation that could conceivably be aided by a selfie-stick, but you're still staring at a screen as you walk rather than taking in the natural surroundings. To alleviate this, I've tweaked the software with an experimental hotspot detection algorithm than plays a sound when something is detected (more below).

Once something is sighted, there is the cumbersome step of trying to look through the binoculars while holding the device:
which seems to work pretty well, but quickly becomes cumbersome when I need to then pull out my camera for a shot. (On my to-do list is to fashion some kind of holster to hold the Therm-App device.)

Software Development


After my initial dabbling with the device, I decided to get the SDK and see if I can't address some of the shortcomings of the device and of the "Basic" app. I'll chronicle here some of the experiments I've done on this front. As you peruse this blog, keep in mind that some of it will be using my experimental app. If you would like a copy, please contact me. I'm happy to share my app for free, but have little time for tech support or proper documentation or making cosmetic improvements.

Background Filtering


The most significant change so far is background filtering, which allows one to take a snapshot (of a blank surface) to serve as a background filter that is subtracted from every frame from the device. I have found that this cleans up a significant amount of noise, especially when the device is used in cold temperatures.


A group of people gathered around a campfire in the woods. The image below is raw while the image above is filtered by a background frame.

Hotspot Detection


I've also started trying various algorithms to detect hotspots, and will play an alert sound when a hotspot is detected. This allows me to walk around pointing the device variously without staring at the screen the whole time.

Color Palettes


Next up on my to-do list is to experiment with different color palettes. In particular, I want a palette that only shows hotspots in bright (yellow), but stretches the remaining temperatures across a wide spectrum of medium-to-low brightness hues, to maximize contrast and detail (thus hopefully to facilitate identifying "landmarks" to help locate the detected hot subject).

8 comments:

  1. Hi Suan.
    I am also a birding utilizing Therm App.
    You it is very want source code of the application that was developed using the SDK. Is it possible to let send the source code to me?

    ReplyDelete
    Replies
    1. You can download a debug apk and/or a source tarball from here:

      http://suan-yong.com/ir/therm_wild.html

      Delete
  2. I am currently working on species reproduction hypogeous character (Tropical shearwather, White-tailed Tropicbird breeding ground in karst cavities) on the island of Europa, in the Mozambique Channel. The nests are not accessible and given the large number of vacant cavity (many favorable sites) we have trouble locating the active breeding sites. Do you think the Therm-App extension can identify the presence of individuals through the karst. This information would allow us more to know the size of the population (which we estimate at this time indirectly with individuals counting the output of colonies) to assess reproductive success later using a burrowscope .

    Thank you in advance for your answer.

    Regards,

    Alexander Laub

    ReplyDelete
    Replies
    1. Having never really looked too carefully at cavities in rocks, I really don't know if it would be effective. Here are some observations that may be informative.

      First, the sun's warmth is overwhelming, so if anything has been exposed to the sun, the ThermApp becomes almost useless for detecting animal warmth. After a sunny day, a tree trunk remain warm for several hours into the night, during which animal warmth within cavities are not easily discernible. I have not tried going out late at night to determine how long it takes for the tree to cool down (probably depends on ambient temperature). If I had to guess, I'd say the effect would be worse for cavities in rocks, which can retain heat for much longer (usually, when scanning the forest floor, rocks appear warmer than leaves and other organic matter). And I would guess animal warmth, especially well-insulated birds, would not affect the temperature of rocks as much as tree trunks.

      Anyhow, with the impact of the sun, the only chance it might work would be either on an overcast day, or possibly early in the morning before sunrise, after the rocks have had a chance to cool down overnight.

      Delete
  3. Hi Suan,
    Thank you for posting your hard-won information on the subject of thermal imagine with wildlife. I would be grateful for your opinion on the Therm-app vs the Therm-app HZ. I need a thermal imaging device for detecting invasive species (rabbits and rats) and seabirds on a subantarctic island. The HZ seems like the suprior choice but I'm concerned about traveling internationally with a 25Hz device. From what I understand, the 8.75Hz basic Therm-app can be taken across borders without any difficulties. But, is it good enough for detecting small mammals and birds? I'd appreciate reading your thoughts on the matter. Thank you.
    cheers,
    peter

    ReplyDelete
    Replies
    1. If it's simply for detecting warm subjects, the 8.75 and 25 are equally effective. The real benefit of the latter is smoother video, if you want to record something nice to look at. The 8.75 video looks very jerky.

      FWIW, getting the bigger 35mm lens will do a lot more for improving detection than upgrading to the 25Hz.

      I made these three comparison videos some time ago:

      https://flic.kr/p/KvaN9N
      https://flic.kr/p/KPyw5k
      https://flic.kr/p/KTwr1C

      Delete
    2. Thanks, that is very helpful. Smooth video would be nice but it's not essential for me. It's more important to have something I can safely travel overseas with. And from your videos I can see the advantage of getting the 35mm lens, too. Thank you again.

      Delete