Experiences Using Cameras to Film Pests in Titirangi, Auckland - David Blake

David Blake is a semi-retired property investor who now likes to spend his time trapping pests and planting native trees. From time to time he helps out by volunteering at The Cacophony Project doing some filming and occasionally contributing to parts of the software.

Approximately two years ago, my wife and I woke in the middle of the night to some scratching sounds. At first, we thought there was a mouse in our bedroom, but we heard it go from one side of the ceiling to the other... and this repeated every night from then on. We had rats in the ceiling. And the noises just got greater and more frequent.

We live in a bush-clad suburb of Auckland called Titirangi.  We had been there 10 years and never heard anything.  I decided to put some rat bait stations around the outside of our house and poison baits in the ceiling.  This lead to two outcomes: 1) Dead rats in the ceiling, which I can tell you, is horrid and 2) Heaps of bait being taken from the bait stations outside. This went on for six months!

So I decided to put the bait stations away, put out traps instead and train cameras on them to see what on earth was going on. Are there really rats running around all over the place eating whatever they can? I wondered. It turns out, yes.  

I bought two Bushnell trail cameras. Ones that can take still images and videos. You can just buy these online through eBay's US site. It costs approximately NZ$230 to land them here.

So I put batteries in the two cameras and put them out in the bush in areas I thought rats might frequent. I trained each camera on a Kness Big Snap-E Trap. These are good traps that you can buy online for $7 or so. Don't buy the cheaper ones from hardware stores, they often don't work.

What I saw surprised me; I was seeing rats all the time. Usually one at a time, but often two or three at a time. On the ground, in the trees, everywhere. I also saw the odd mouse and possum. I caught a lot of rats. 60 in the first year.  I wanted to film this, but I found it difficult to achieve.

I even started my own YouTube channel as a way to easily share these videos

What I found was that the batteries in the cameras drained quickly. On the manufacturer's site it says the batteries will last a year. I imagine that might be the case if the camera takes one photo every few days. But with my cameras taking dozens of videos and photos each night, the batteries lasted about a week. I wanted to take videos because they show so much more than photos. Like how the rats move, what direction they come from and so on.

But this is very draining on batteries. The cameras have to operate a PIR (this is the same thing that is in the sensor that detects movement in home security systems many people have) and then if it's night time, illuminate the scene with infrared light and then record the action.  I was finding that I was having to check battery levels constantly, change batteries often (they have to be near full power to record video) and worst of all, I was missing a lot of what I was trying to film, which was frustrating.  And of course, good quality batteries aren't cheap.

Other problems I found with battery powered trail cameras:

1) With recording videos, you can't leave it recording all the time, because the batteries would quickly become drained. So what the cameras do is they start recording when movement is sensed. And you set a recording duration and then an interval for the camera to wait until it starts recording again.  So I might set the camera to record for 30 seconds and then wait 10 seconds before recording again, if movement is still detected.  What I found with this is that I almost always missed the point at which the rats got caught in the traps. They would seem to get caught during that 10 second window, or before the 30 second recording period. This was very frustrating. I played around changing settings e.g. recording for one minute, with a three second (which is the shortest you can set) gap, and other settings like sensor sensitivity, movement sensitivity etc. But I still found I was missing a lot of footage.

2) I ended up filming a lot of things I wasn't interested in. Often I would have the cameras quite close to the ground because my traps would be mounted on a tree or fence, say 200mm above the ground. Blackbirds and thrushes hop around all over my section (and everyone's section) so I would get 20 videos each morning of birds digging up worms etc. This, of course, is a waste of time and batteries. Later on I bought a slightly more expensive camera which has a 'night mode' which was very good - it simply didn't take any videos during the day. And since rats are almost never sighted during the day, I didn't miss anything worthwhile.

3) Sometimes the cameras just didn't detect anything. Quite often when checking the cameras each morning, I'd find a rat in a trap, but no footage on the camera of it. To be fair, these cameras are made for filming large animals like deer, pigs, bears etc. And at distances of say 10 to 30 metres.  They are not made for filming small animals like rats, or even animals the size of possums. So they do remarkably well given that, but still, the problem remains that they miss some of the 'goings on' I'd like to film.

4) The cameras get set off by random things such as a lot of vegetation movement. So on windy nights I'd often get lots of videos of nikau palm fronds waving. And sometimes what I think happens is a pest runs by, the camera triggers but is too slow, so that by the time it is recording the pest has already gone. The cameras advertise very fast trigger speeds, but I feel these may be exaggerated.

5) Setup. I found that I would have to spend a lot of time either strapping the cameras to nearby trees or to wooden stakes I'd hammered in to the ground, hopefully angling it correctly towards what I wanted to film. But often the next morning I'd see a rat caught in a trap and then when I looked at the footage, I'd find the camera was pointing too high or to the left, for example. So I had missed the thing I was trying to film. Some more expensive cameras can send an SMS message to you showing you a frame of what the camera is looking at. I would try and buy one of those next time.

6) More on setup: To record successfully you have to have a) Fresh batteries, b) An SD card in the camera and c) The camera turned on! And, of course, if you are wanting to trap your critter, then the trap set! This all sounds straightforward doesn't it? But I'd often find that I missed one of these things. So I'd come back in the morning to find that I had everything nicely prepared but hadn't turned the camera on. Or I'd left out the SD card. And of course sometimes I got crystal clear footage of rats stealing peanut butter out of my unset traps :(

7) Water ingress. The camera has to be opened to put batteries and SD cards in and out. And if you're doing this on a wet, humid day, a certain amount of moisture gets in. Think about four guys getting into a car when it's raining. The windows steam up with all the introduced moisture.  So I found sometimes the lenses of the camera would get foggy.  Therefore I started doing two things: 1) Mounting an upside down ice cream container above them, to keep the bulk of the rain out and b) Putting the cameras into the hot water cupboard in the open position, now and again, to dry them out.

The good things about using battery powered trail cameras:

1) No running power cords. This is pretty obvious. Running power cords around your section, let alone remote areas is just out of the question for all places other than a few meters from your house.

2) The footage that I did capture was of surprisingly good quality. The videos and images were generally crisp and clear. I could clearly see what the rats and other animals were doing. I was able to see, for instance, how long a rat takes to die in a trap - because the camera films when they get caught and stops filming when they stop wriggling. We generally only see them dead when we check the trap.  But they don't die instantly of course, it can take from three seconds to 30 minutes to die. One rat took around 6 hours to die... I know this because the trap was on the ground and a hedgehog came up to it and bit the rat and it moved! Because the footage is so detailed, when the camera is tripped by, say, a rat, you can also see things like slugs and snails slowly eating the peanut butter out of traps, which is annoying, but at least tells you why your bait is disappearing!

Here is a still image from one of the trail cameras. It shows a rat caught in a snap trap and another one coming to investigate.

3) The pests I filmed were not perturbed by them. The cameras have to shine an infrared light in front of them to be able to film the pests. Conceivably, the pests could see this, get frightened and run away. But they didn't seem to do this at all. Of course, I can't be sure that some of them didn't get frightened and run, but I did film hours and hours of pests right in front of the cameras oblivious to them.

Here is a shot of a possum right next to a Timms trap.  It was not worried about the camera at all.  This possum ate a piece of pre-feed apple outside the trap, then left, not even investigating the apple on the trigger mechanism inside the trap. Which says something about the effectiveness of the trap.

For a few weeks I've been using one of the thermal cameras we are working with here at The Cacophony Project. It is dubbed a "Cacophonator" because eventually, it won't just take pictures...

This is quite a different beast. It takes those thermal kind of images and videos everyone has seen where cold things are blue and warm things are yellow, orange, red etc. See the technology for a more detailed description of this camera. 

The good things about the thermal camera:

1) It is very sensitive. It only needs a very small temperature change to trigger. The camera is always on and we use software to detect movement, and this is adjustable. It appears to capture all animals (anything that is warm) within a range of one to 30 metres (I haven't tested further as yet). What you get is video of an animal moving with the background being all blue/black and the animal highlighted in the warmer colours I mentioned above. It films mice, people and everything in between.

To test the effectiveness of the thermal camera at detecting the presence of rats, the Cacophonator was set up next to a Bushnell trail camera.  Both cameras were directed at a GoodNature chew card approximately two metres away.  They were set up so that they were at the same height from the ground and the same distance from the chew card. A couple of things were noticed:  Firstly, whilst both cameras recorded a rats moving past, the thermal camera recorded more instances of this.  The trail camera did not activate and record as often as the thermal camera.  And secondly, the chew card remained untouched even though the rat passed within approximately half a metre of it.
So this would tend to suggest that the Cacophonator may be better at detecting the presence of rats than both  GoodNature chew cards or Bushnell trail cameras.

2) At this stage the Cacophonator is powered by a mains cable. This is good in that I don't have to worry about batteries at all, which is a huge relief. But of course, as this device is deployed to more remote areas, a battery pack will become a must.

3) The footage is uploaded to our API server where they can be filtered, viewed and annotated. So the footage can be seen at any time, without the need to physically go into the bush and get it via the SD card. This saves a huge amount of time as it allows me to sit down in front of my PC with a coffee and check to see if anything interesting was filmed overnight. If there's nothing there, then I know I don't have to go and check the trap. If there is, I can see what it is, what time it was filmed, what our algorithms classified it as and so on.

This is achieved in one of two ways; the Cacophonator either connects to a WiFi network, if one is nearby, or uses a 3G mobile connection to connect to the internet and upload footage that way. In the future The Cacophony Project will be testing other communication methods such as LoRa which may be more suitable for very remote locations.

4) This connectivity means no SD cards, and no forgetting to put them in the camera, or getting back to base and realising you forgot to change the SD card in one or more cameras.

5) The thermal footage can be used by our algorithms to classify animals. Being able to identify what pest or non-pest is being viewed by a machine rather than a person brings enormous benefits. One of which is saving time , so that the time a person has to spend looking at footage is much reduced.

6) The Cacophonator can be set to various modes such as Night, Day or Both. Once again, this will save unnecessary footage being taken. For instance, when a person goes to check on the camera (or perhaps a trap near it) to change its settings.

7) We can remotely change a Cacophonator's settings. Since the Cacophonator is also a computer (a Raspberry Pi) we can view and change its settings from any device, anywhere. When deployed in the field, this will save a lot of time and hassle. We will be able to see if the Cacophonator has connectivity, what its battery levels are, how much disk space and memory it has, what recording mode it is in and many other things.

Problems I have found using a thermal camera (Cacophonator) 

1) I can't see the background very clearly. Because everything other than the animal is cold, I can't see the trap(s) I have set. So, I can't clearly see if the pest is entering a live capture cage, or say, sniffing a bait. If the pest goes behind an object like a tree trunk it kind of 'disappears' for a second or so. But the latter happens with any camera, of course.

2) Mains power: As I said above, the Cacophonator is currently mains powered, but this will change and is a temporary issue while we develop the device.  The reason for the high power consumption is that it is a computer running and trying to detect predators in a much more sophisticated way – see 1) above. At this stage we have battery packs which allow the device to be powered by batteries, but only for 5-6 days. Over time we hope to extend this time outwards.

3) Connectivity: In more remote locations something other than WiFi or 3G connectivity will be needed to be used. Again, this is being investigated.

The Cacophonator removes most of the problems with battery powered trail cameras, and with the way that technology tends to get better and cheaper over time (as espoused by things like Moore’s Law) we feel the power and connection issues will disappear over time.

The project is an open source one, and so we are finding that along with our very dedicated and competent staff, volunteers are helping in many areas. And because the Cacophonator has a computer at its heart, in the near future it will allow the rapid testing of sound lures, enhanced elimination methods and link to all sorts of analytics showing what is going on.

I hope this gives everyone a good feel for the use of cameras to detect and film pests in the bush (in fact, anywhere) in New Zealand.