Very impressive Kim! It actually took me a minute to find them in the first picture! Are they baboons (if they aren't then this is definitely proof that I shouldn't creat any training datasets!)? Can't wait for the thermal tests!!

Cheers,

Rob

Thank you for sharing! 

Hi Brett,

I am interested in developing really low cost multi channel underwater acoustic recorders.  Can you tell me a bit more about the board and stuff you were using for yours? You can get me at chris@wilcoanalytics.org.

Thanks,

Chris

Cool thread!

I will be testing Reolink Wi-Fi cameras in combination with solar powered TP-Link long range Wi-Fi antennas/repeaters later this field season for monitoring arctic fox dens at our remote off grid site in Greenland. The long range Wi-Fi antennas are rather power hungry but with sufficient solar panel and battery capacity I am hopeful it will work. 
I am looking forward to explore the links and hints above after the field season. 
Cheers,

Dear team,

I am new to WildLabs. It's great to find all the interesting and informative discussions.

We are exploring deploying "connected" camera traps in wetlands (2 hectares for the pilot, 300 hectares for the next phase) and our goals are:
- Set up the camera traps to upload images (via FTP or HTTP POST) to cloud storage every 5 minutes
- Analyze the images using AI to remove all false positives (images without animals)
- Deploy off-the-shelf camera traps that support WiFi (preferred) or 4G cellular

Questions:
a. Any suggested camera traps that support WiFi (preferred) or 4G cellular?
b. What WiFi infrastructure is recommened to connect camera traps at 200m to 500m distances?
c. Does anyone have experience using MANET (mobile ad-hoc network)?

Cheers,
Andre

Hi Ben,

I would be interested to see if the Instant Detect 2.0 camera system might be useful for this.

The cameras can transmit thumbnails of the captured images using LoRa radio to a Base Station. You could then see all the captured images at this Base Station, as well as the camera's battery and memory information (device health). In addition, you could also change camera settings from the Base Station so you would not need to reclimb the trees to change from PIR sensitivity high to medium for instance.

The Instant Detect 2.0 cameras also have an external power port so a cable could be run to the ground to a DC 12V battery for long term power.

If you wanted to, you could also connect the Base Station to the Cloud using satellite connectivity, so that you can monitor the whole system remotely, but this will incur more cost and power usage of the Base Station.

I'd be keen to hear your thoughts,

Thanks,

Sam 

Hi Karen. 

It's actually possible that it might not necessarily be the high heat that is causing problems with the batteries. Lithium should be quite durable for the temperatures experienced inside the enclosure. Instead, the problem could likely be condensation. In an enclosure, if the sun shines directly on a sealed enclosure, the enclosure can internally heat up, evaporating any moisture in it and raising the dew point. If the external temperature then cools, the internal enclosure air that comes into contact with the cooler walls of the enclosure can form condensation droplets. If you've ever left a glass jar in the sun, you'd see this phenomenon. 

The problem is that if any condensation gets on the internal camera circuit board, its possible that it can potentially conduct enough current to increase the discharge rate of the batteries or in some cases, cause the whole device to go down. We've seen this happen in some of the field devices we deploy and have since protected critical circuitry against condensation.   

To avoid this situation, I'd recommend trying to keep your camera traps out of direct sunlight and putting a lot of dessicant inside before sealing them. Not sure if this solves your problem, but hopefully it might help. 

Im a biologist up in Alaska and have a couple of options that might work for you. We have a time lapse camera we have developed for 'peering' into the nest cup of shorebirds and passerines (focus is set at about 6 inches with a 150 deg lens) and a small mammal camera is designed to sit on a 3 gallon bucket and look downward (about a 11 in focus). I'm curious how these might perform in this scenario so am happy to send a couple for you to try out. Please shoot me an email at my work email address if you are interested: christopher_latty@fws.gov

Support for two cameras (or more) can be done by using two USB powered cameras if you have the right platform. Then in principle you can place each camera up to 2m from the processor.

I’ll will be experimenting with such a setup as soon as my USB global shutter camera arrives. Global shutter camera work really well with the AI computer vision that my system uses. Quite a different power category than microcontrollers though.

Currently I’m testing a Raspberry Pi based camera system that is capturing images and video, scanning 3x cameras with ai triggering.

great idea!!

@cmwainaina please take a look

I would separate your question into two, re: (a) platforms and (b) AI.

Re: platforms... 

One of the advantages (the main advantage?) of working with a cloud-based system is that the storage and data management issues are handled for you.  So if a cloud-based system is on the menu for you, check out any of the cloud-based systems I list here (Wildlife Insights, Agouti, TrapTagger, TRAPPER, WildTrax, Animl, etc.).

But cloud-based systems are not a panacea, and there are lots of reasons folks choose to work locally.  Often working locally is preferred due to bandwidth, data provenance, or being absolutely sure that your images are archived somewhere you can still access them in five years, but also in general when it comes to pushing lots of images in front of your eyes quickly, nothing will beat a local tool, and among local tools, Timelapse is still the most widely used AFAIK and the fastest for most users.  I'll get to this below, but AI might just not help you in this case, so my best recommendation is to invest in really getting to know all of Timelapse's bells and whistles, especially the "quick paste" tool, and all of the keyboard shortcuts.  If you're using the mouse to navigate, you've lost the battle already, and 30 minutes learning keyboard shortcuts is worth a week of tinkering with AI.

FWIW in general, no matter what system you use, you'll have to do something with .csv files in R, and overall I wouldn't expect that to be more or less difficult depending on the system you choose.

Re: AI...

@carlybatist is correct, I loooooove difficult MegaDetector cases.  Based on your description, it may be the case that small rodents are too difficult, but don't give up hope quite yet; I have a bag of tricks I try in these more difficult cases.  I will reach out by email to ask you for some images, and if any of the tricks work, I will post back to this thread.

It also couldn't hurt to try a couple other systems in parallel, just to see what happens, but I'm not optimistic that any other systems will work if "vanilla MegaDetector" doesn't work, in part because the systems that I think are most relevant themselves depend on MegaDetector.  But FWIW, the two systems I would try - even if cloud-based systems are not on the menu for you - are Wildlife Insights and Animl.  There are other systems out there as per above; I mention Animl specifically because it has grown up around small mammal monitoring on islands, so, maybe a better-than-average shot of doing something reasonable in your case.

IMO the best path to pursue is "Dan's bag of MegaDetector tricks", but I want to set expectations: small nocturnal rodents remain difficult for AI and the odds are not in our favor.

I'm being quite daring by making statements on a forum that is frequented by Dan Morris (obviously in deference to his extensive knowledge on the subject :-) ), but here goes. Likely the version of megadetector that you used was using YoloV5. YoloV5 came out a similar time to YoloV4, which from memory was around end 2019 time frame. Again from memory I think the performance of Yolov5 was similar to that of YoloV4. The versions that came out after this, (Somewhat in order) Scaled Yolov4, Yolor and then yolov7, yolov6 (Yolov8 somewhere in there) are much better at matching smaller objects and in darker conditions. Dramatically so.

PyTorch Wildlife will be releasing a beta of a version that will be using YoloV9. This could possibly help with your problem.

Co-incidentally, I'm playing with megadetector now for detecting and recording mice. This was from this morning:

This reminds my of this preprint:

FoxMask: a new automated tool for animal detection in camera trap images

https://www.biorxiv.org/content/10.1101/640037v1

We evaluated background subtraction as part of our pipeline for Zamba Cloud a few years ago. It's going to depend a lot on your backgrounds, but we found that for jungle/forest scenes in videos there was too much movement/noise in the backgrounds for the usual methods to add substantial benefit to the accuracy of our classifications.

To classify frames, we run a very fast distilled version of megadetector to select frames from videos that are likely to have animals and then classify using a CNN. We found this to be much more computationally efficient with respect to accuracy gain versus RCNNs or some dual stream approaches that try to account for movement with something like optical flow.