discussion / Camera Traps  / 25 October 2019

Virtual Meetup Discussion: Camera Trapping

Hi Wildlabbers, 

I'm looking forward to kicking of season 3 of our virtual meetups with you all next week! We've already had 250+ registrations from 54 countries, so it looks like this is going to be a popular one. If you want to join, you need to register here.

This session we're going to be joined by @Rolandisimo, @sarabeery and @Soseccombe . Ahead of the meetup I've been sorting through all of your comments and questions raised during registration. The key discussions it looks like you want us to cover include: 

Best Practice

  • 10+ on Survey Design and general best practice for setting up camera trap programs  (Ecuador, USA, UK, Kenya). Advise looking at links I include at the bottom of this post. Examples include: 
    • How to effectively use camera traps to determine density of few individuals that are vastly distributed? (Kenya)
    • How can I improve my camera deployment strategy to answer the question of meso-predator interactions in my study area? (USA)
    • Best protocols for long-term population monitoring (USA)
    • How do you modify a camera trapping experiment for different species? (USA)
    • How easy is it to use camera trap data to assess activity/density/abundance? And how easy (or necessary) to tease these apart? (USA).

Hardware

  • Improvements
    • What technological advances are over the horizon? (Aus, USA, Peru)
    • AI on the edge - can AI on the camera to distinguish between human and animal/wildlife? (South Africa)
    • Is anyone looking at increasing selectivity for real targets vs background vegetation ? (Botswana)
    • Reasons behind and prevention of common camera malfunctions in the tropics (excessive triggers, drained batteries) (USA)
    • Can we develop camera traps for reptiles and amphibians? (Pakistan)
  • Best practice
    • To bait or not to bait? (Canada)
    • What are the best and most effective camera traps for a range of common scenarios? (UK, Bhutan)
    • Are there any good remote download technologies (using come combination of bluetooth, WiFi and/or LoRa) available? Low cost? (NZ, Brazil, UK, USA)
  • Looks like we're going to be joined by quite a few members who are designing/developing camera trap hardware (both commercial and open source) and interested in seeing what possible improvements they can make or share. They are also keen to add to the knowledge base.  (Netherlands, India, Australia, South Africa, Japan). 

Data Management + Processing (AI/Machine Learning)

  • 20+ questions from USA, Kenya, UK, Nepal, India, Argentina, Puerto Rico, Malaysia. are best summed up by the member who asked: Where are we at with AI autotagging of photos? 
    • Machine learning pros and cons? Best way to train? (UK)
    • How many people are using AI to support processing and can we rely it to report as accurately as a human? (Czech Republic)
    • What are the best current tools for CT data handling and analysis (UK, Malaysia)
    • What's the latest progress with Wildlife Insights (Myanmar)
  • What metadata standards and repositories are people using to manage their images? (UK)
  • How can we update Camera Trap Metadata Standard to include information generated by machine learning systems? (USA)

Collaboration + How do I get Involved? 

  • I use AI techniques for unique identification of tigers in camera trap images as well as monkeys captured in hand held devices. How can people from different backgrounds can come together to work together to solve these crucial problems? (India)
  • Central database? (germany)
  • How to do shared research and cost sharing after deployment? (USA)
  • Is there any open project to build camera traps at lower costs? (Argentina)
  • EE data/image processing skills, want to get involved in conservation - where are some volunteer opportunities? (USA)
  • We need versatile, buildable designs using real cameras. CHDK is a great leap forward. (Aus)
  • What is the best way for local communities to be involved in camera trap data collection? (Namibia) 

Given all of the questions around best practice, there are a few resources that might be useful: 

Finally, our camera trapping group here on WILDLABS is a fantastic resource for you to access support for your camera trapping work on an ongoing basis. If you are looking for advice about hardware, survey design, or how you can get involved - please join and post there. 

See you all next week!

Steph

 




Hi everyone,

I don't want to bore you all anymore than I already have today but I thought I would point out this thread of some results I found when I first started comparing camera traps. All the models tested will be a bit old now but I think it helps you see the differences between cellular and non-cellular camera performance.

https://www.wildlabs.net/community/thread/470

Thanks,

Sam

Sam Seccombe - I was very interested to hear that pictures can be sent by LoRa since my understanding was that its data packages are tiny - just enough for a time and a GPS fix from a tracking collar. Are there any good links on transmitting (much, much) larger data streams ?

Nobody talked about videos, or about using camera traps to record behaviour. Seeing that Microsoft is involved made me wonder if they might find a way of making their Explorer video thumbails show a frame from a couple of seconds into the video rather that from a later (and variable) frame, by which time the animal that triggered the camera has walked out of shot so that to sort false from genuine triggers means opening all the videos which do not have an animal in their thumbnail.