discussion / Camera Traps  / 8 May 2019

Advice and construction of camera traps.

Howdy geniuses.

I'm writing from Australia with an interesting one for you.

6 months ago a group of us started a project that got initial seed funding from Google.org. it's called Desert Ears. A simple one pagers here - https://www.environskimberley.org.au/https_www_environskimberley_org_au_desert_eyes_ears

Our idea was a simple one, to develop methods of democratising ecological monitoring for indigenous ranger groups and to increase what we know about the Australia natural landscape.  

The first part was easy, find a way to automatically categorise images with AI…ha!  But …we found three different options. Now we’re starting to look at data collection to use with this software.

The easiest and simplest thing to do would be to purchase cameras and just make them work for our project. But I don’t like the simplicity of that, plus it’s not actually solving our problem: Australia is inaccessible, huge and has no infrastructure to send and receive data remotely- well not affordably anyway. For instance, our trial site is three days drive. No roads, no nothing. Just glorious sandhills and spinifex plains. In terms of connectivity we’ve started conversations with the Australian owned satellite internet service NBN sky muster to which we are expecting them to support the project. Now we just need a camera that:

-Is cheap

-Runs on solar

-Can process images locally and;

-Then ‘pushes’ images with animals in them to the cloud.

Anyone out there interested?

Toby Barton

Walkes

E - toby.w.barton(AT)gmail | Desert Eyes | Fire




Hi Toby,

At my work we're developing exactly that kind of tech, specifically in mind for wildlife detection. We ran a small trial in South East Asia last year and are planning on running a more advanced one this year.

I think there's a very good match with your needs: smart cameras, remote comms (e.g., satellite), and low power/solar power :)

We'd love to have a chat if you're interested.

Let me know.

Ivan

Hi Andy,

Yes you're on the money regarding processing. We will be using batteries and solar to run the units. likely our first prototypes will be just on a stake in the ground with a battery, solar, housing for computer and camera and a weather station on top. All processing and sending data will be done in the day whilst the sun is shining.

we're lucky that it looks like we've stumbled into a partnership with the NBN skymuster satellite services in Australia that will help us send and receive data from anywhere in Aus. reasonably cheaply whilst we're in the prototyping phase.

We hope that through the prototyping phase we’re able to work with a group called classify me to develop AI models that can identify specific species. i.e. a dog/cat/camel/horse etc. then more specific Aus. mammals like a quoll.

Once we’ve got a model like that we can put out cameras that just look for a specific species and send small bit rate data to notify us. This is the model that we want to work with currently https://classifymeapp.com/

here is there first publication and a test camera that theyre using.

 

https://www.wool.com/globalassets/start/about-awi/publications/beyond-the-bale-75-june-2018.pdf