We saved all images to a harddrive & to the cloud. We used Microsoft’s MegaDetector for object detection in each image & Timelapse2 to review & manually tag images. Not sure if our download/folder system was the best approach but it worked for us with 120 cams running for 3 yrs.
December 4, 2024 at 4:08 PM
We saved all images to a harddrive & to the cloud. We used Microsoft’s MegaDetector for object detection in each image & Timelapse2 to review & manually tag images. Not sure if our download/folder system was the best approach but it worked for us with 120 cams running for 3 yrs.
📊 We showed that CT-DS and REM both provide reliable estimates when an algorithm (#MegaDetector,
@sarameghanbeery.bsky.social @doctorofrock.bsky.social) is used to filter blank images, and CT-DS can even be used with an automated taxonomic classifier (#WildlifeInsights) with minimal bias.
@sarameghanbeery.bsky.social @doctorofrock.bsky.social) is used to filter blank images, and CT-DS can even be used with an automated taxonomic classifier (#WildlifeInsights) with minimal bias.
December 3, 2024 at 2:58 PM
📊 We showed that CT-DS and REM both provide reliable estimates when an algorithm (#MegaDetector,
@sarameghanbeery.bsky.social @doctorofrock.bsky.social) is used to filter blank images, and CT-DS can even be used with an automated taxonomic classifier (#WildlifeInsights) with minimal bias.
@sarameghanbeery.bsky.social @doctorofrock.bsky.social) is used to filter blank images, and CT-DS can even be used with an automated taxonomic classifier (#WildlifeInsights) with minimal bias.
The auto AI on StealthCam is great
July 9, 2025 at 5:44 PM
The auto AI on StealthCam is great
NEW PAPER 🔍
How can we best quantify human-wildlife activities in protected areas?
#Cameratraps combined with object detection models (#MegaDetector) show promise! With our friends in Bayreuth we evaluated how this method supports long-term monitoring:
doi.org/10.1002/rse2.3…
How can we best quantify human-wildlife activities in protected areas?
#Cameratraps combined with object detection models (#MegaDetector) show promise! With our friends in Bayreuth we evaluated how this method supports long-term monitoring:
doi.org/10.1002/rse2.3…
February 3, 2025 at 10:29 AM
NEW PAPER 🔍
How can we best quantify human-wildlife activities in protected areas?
#Cameratraps combined with object detection models (#MegaDetector) show promise! With our friends in Bayreuth we evaluated how this method supports long-term monitoring:
doi.org/10.1002/rse2.3…
How can we best quantify human-wildlife activities in protected areas?
#Cameratraps combined with object detection models (#MegaDetector) show promise! With our friends in Bayreuth we evaluated how this method supports long-term monitoring:
doi.org/10.1002/rse2.3…
‼️🔴 Our results suggest that popular algorithms for animal detection such as MegaDetector can be safely integrated with CT-DS and REM, and that final density estimates are reliable. Moreover, CT-DS showed to be robust even when taxonomic classifier accuracy was as low as 50%.
December 3, 2024 at 3:26 PM
‼️🔴 Our results suggest that popular algorithms for animal detection such as MegaDetector can be safely integrated with CT-DS and REM, and that final density estimates are reliable. Moreover, CT-DS showed to be robust even when taxonomic classifier accuracy was as low as 50%.
With hundreds of cameras and millions of photos in our Flow Photos Explorer, we get some cool wildlife pix (screened by megadetector). See the owl?
Nerd alert - we can also tell you within 15 min when the tree top of the stump the owl is sitting on fell. That's got to be valuable to someone, right?
Nerd alert - we can also tell you within 15 min when the tree top of the stump the owl is sitting on fell. That's got to be valuable to someone, right?
November 27, 2024 at 8:45 PM
With hundreds of cameras and millions of photos in our Flow Photos Explorer, we get some cool wildlife pix (screened by megadetector). See the owl?
Nerd alert - we can also tell you within 15 min when the tree top of the stump the owl is sitting on fell. That's got to be valuable to someone, right?
Nerd alert - we can also tell you within 15 min when the tree top of the stump the owl is sitting on fell. That's got to be valuable to someone, right?
📊 We showed that CT-DS and REM both provide reliable estimates when an algorithm (#MegaDetector,
@sarameghanbeery.bsky.social @doctorofrock.bsky.social) is used to filter blank images, and CT-DS can even be used with an automated taxonomic classifier (#WildlifeInsights) with minimal bias.
@sarameghanbeery.bsky.social @doctorofrock.bsky.social) is used to filter blank images, and CT-DS can even be used with an automated taxonomic classifier (#WildlifeInsights) with minimal bias.
December 3, 2024 at 3:26 PM
📊 We showed that CT-DS and REM both provide reliable estimates when an algorithm (#MegaDetector,
@sarameghanbeery.bsky.social @doctorofrock.bsky.social) is used to filter blank images, and CT-DS can even be used with an automated taxonomic classifier (#WildlifeInsights) with minimal bias.
@sarameghanbeery.bsky.social @doctorofrock.bsky.social) is used to filter blank images, and CT-DS can even be used with an automated taxonomic classifier (#WildlifeInsights) with minimal bias.
‼️🔴 Our results suggest that popular algorithms for animal detection such as MegaDetector can be safely integrated with CT-DS and REM, and that final density estimates are reliable. Moreover, CT-DS showed to be robust even when taxonomic classifier accuracy was as low as 50%.
December 3, 2024 at 2:58 PM
‼️🔴 Our results suggest that popular algorithms for animal detection such as MegaDetector can be safely integrated with CT-DS and REM, and that final density estimates are reliable. Moreover, CT-DS showed to be robust even when taxonomic classifier accuracy was as low as 50%.
Results 1: 21 day experiment, time-lapse cameras serviced every few days.~410,000 images, 97% were time-lapse series.
To remove images w/o animals, we first used Microsoft’s #Megadetector, which cut down the number of images needing manual tagging by ~90%!
#ABMIatWork
5/7
To remove images w/o animals, we first used Microsoft’s #Megadetector, which cut down the number of images needing manual tagging by ~90%!
#ABMIatWork
5/7
February 26, 2025 at 4:47 PM
Results 1: 21 day experiment, time-lapse cameras serviced every few days.~410,000 images, 97% were time-lapse series.
To remove images w/o animals, we first used Microsoft’s #Megadetector, which cut down the number of images needing manual tagging by ~90%!
#ABMIatWork
5/7
To remove images w/o animals, we first used Microsoft’s #Megadetector, which cut down the number of images needing manual tagging by ~90%!
#ABMIatWork
5/7
SpeciesNet, 카메라 트랩을 위한 AI 기반 야생동물 분류 모델 (feat. Google)
(by 9bow님)
https://d.ptln.kr/6277
#google #megadetector #wildlife #speciesnet
(by 9bow님)
https://d.ptln.kr/6277
#google #megadetector #wildlife #speciesnet
SpeciesNet, 카메라 트랩을 위한 AI 기반 야생동물 분류 모델 (feat. Google)
SpeciesNet 소개 Google이 야생동물 보호 및 연구를 위한 AI 모델인 SpeciesNet을 공개했습니다. 야생동물 모니터링을 위한 카메라 트랩(camera trap)에서 촬영된 방대한 이미지 데이터를 자동으로 분석하는 AI 시스템입니다. 기존에는 연구자가 직접 수백만 장의 이미지를 확인해야 했지만, 이제 AI가 이를 자동으로 분류하고 처리할 수 있습니다. 야생동물 보호 및 연구는 수많은 이미지 데이터를 처리해야 하는 작업입니다. 연구자들은 카메라 트랩을 이용해 동물의 행동과 개체 수를 모니터링하지만, 그 결과 생성되는 이미지 데이터가 너무 많아 사람이 직접 분류하는 것은 쉽지 않습니다. SpeciesNet은 이를 해결하기 위해 만들어진 AI 모델로, MegaDetector와 SpeciesNet 분류기 두 가지 모델을 결합한 앙상블 시스템입니다: MegaDetector: 이미지에서 동물, 사람, 차량을 탐지하는 객체 탐지 모델 SpeciesNet 분류기: 탐...
d.ptln.kr
March 4, 2025 at 3:42 AM
SpeciesNet, 카메라 트랩을 위한 AI 기반 야생동물 분류 모델 (feat. Google)
(by 9bow님)
https://d.ptln.kr/6277
#google #megadetector #wildlife #speciesnet
(by 9bow님)
https://d.ptln.kr/6277
#google #megadetector #wildlife #speciesnet
No sign-up needed. You can share images via URL 🔗 or Download ⏬ them.
The tool uses same AI models as our Animal Detect platform
-> MegaDetector for detection
-> SpeciesNet for classification
These two models are the best general models out here.
The tool uses same AI models as our Animal Detect platform
-> MegaDetector for detection
-> SpeciesNet for classification
These two models are the best general models out here.
August 4, 2025 at 7:00 AM
No sign-up needed. You can share images via URL 🔗 or Download ⏬ them.
The tool uses same AI models as our Animal Detect platform
-> MegaDetector for detection
-> SpeciesNet for classification
These two models are the best general models out here.
The tool uses same AI models as our Animal Detect platform
-> MegaDetector for detection
-> SpeciesNet for classification
These two models are the best general models out here.
New paper alert ⚠️ Using #AI tools like #megadetector and #birdNET to process camera trap images or audio recordings?
Read our perspective piece for some considerations and guidance on
📊 working with 0-1 confidence scores
🤔 making thresholding decisions
🧑💻 and navigating AI-labelling errors
Read our perspective piece for some considerations and guidance on
📊 working with 0-1 confidence scores
🤔 making thresholding decisions
🧑💻 and navigating AI-labelling errors
Improving the integration of artificial intelligence into existing ecological inference workflows
Artificial intelligence (AI) has revolutionised the process of identifying species and individuals in audio recordings and camera trap images. However, despite developments in sensor technology, m...
besjournals.onlinelibrary.wiley.com
April 14, 2025 at 4:15 PM
New paper alert ⚠️ Using #AI tools like #megadetector and #birdNET to process camera trap images or audio recordings?
Read our perspective piece for some considerations and guidance on
📊 working with 0-1 confidence scores
🤔 making thresholding decisions
🧑💻 and navigating AI-labelling errors
Read our perspective piece for some considerations and guidance on
📊 working with 0-1 confidence scores
🤔 making thresholding decisions
🧑💻 and navigating AI-labelling errors
Just look at this deer!!
Well, I'm 91.4% certain it's a deer.
Otherwise, it's a pig!
Yes, I'm 62.3% certain it's a pig.
I've been identifying camera trap images with the MegaDetector & NZ Invasives image ID models in the EcoAssist app.
While some people […]
[Original post on mastodon.nz]
Well, I'm 91.4% certain it's a deer.
Otherwise, it's a pig!
Yes, I'm 62.3% certain it's a pig.
I've been identifying camera trap images with the MegaDetector & NZ Invasives image ID models in the EcoAssist app.
While some people […]
[Original post on mastodon.nz]
December 21, 2024 at 10:51 PM
Just look at this deer!!
Well, I'm 91.4% certain it's a deer.
Otherwise, it's a pig!
Yes, I'm 62.3% certain it's a pig.
I've been identifying camera trap images with the MegaDetector & NZ Invasives image ID models in the EcoAssist app.
While some people […]
[Original post on mastodon.nz]
Well, I'm 91.4% certain it's a deer.
Otherwise, it's a pig!
Yes, I'm 62.3% certain it's a pig.
I've been identifying camera trap images with the MegaDetector & NZ Invasives image ID models in the EcoAssist app.
While some people […]
[Original post on mastodon.nz]
Do you use megadetector to dp the first sift?
November 17, 2023 at 7:56 AM
Do you use megadetector to dp the first sift?
Many thanks. Feel like we're going this route but have to find our own custom. Megadetector documentation looks very friendly compared to some other platforms we looked at. I do need to identify watercraft from images, so am expecting well need to do some training regardless! Many Ty again!
December 5, 2024 at 2:23 AM
Many thanks. Feel like we're going this route but have to find our own custom. Megadetector documentation looks very friendly compared to some other platforms we looked at. I do need to identify watercraft from images, so am expecting well need to do some training regardless! Many Ty again!
Thanks to the great cooperation with #Microsoft and the use of Megadetector were able to speed up our processing of camera trap pictures. Now we can accurately classify large amounts of data accurately without human intervention
customers.microsoft.com/en-au/story/16…
customers.microsoft.com/en-au/story/16…
Fifty thousand nature shots in two days: The Bavarian Forest National Park is simplifying its wildlife monitoring with Microsoft Azure and AI | Microsoft Customer Stories
It’s a picture of biodiversity: 14,000 species living on more than 25,000 hectares. To safeguard this diversity for the future, camera traps are used to continuously monitor wildlife populations in the Bavarian Forest National Park, central Europe’s oldest and largest forest preserve. The traps capture some 1,000 images a day of anything that moves, such as animals and plants. The images are then analyzed and categorized for research purposes and population surveys. Previously a very time-consuming task involving manual sorting, the same volume of data can now be processed in just two days—thanks to MegaDetector, Microsoft Azure, and automatic categorization by artificial intelligence. As a result, insights are gained faster, paving the way for even more species conservation.
customers.microsoft.com
February 3, 2025 at 10:29 AM
Thanks to the great cooperation with #Microsoft and the use of Megadetector were able to speed up our processing of camera trap pictures. Now we can accurately classify large amounts of data accurately without human intervention
customers.microsoft.com/en-au/story/16…
customers.microsoft.com/en-au/story/16…
Currently, roughly 90–95% of AI usage in biodiversity
and conservation research is simply identifying
a species of interest in gobs of data, says
@sarameghanbeery.bsky.social of the Massachusetts Institute of Technology, co-founder of MegaDetector.
But that promises to change swiftly.
and conservation research is simply identifying
a species of interest in gobs of data, says
@sarameghanbeery.bsky.social of the Massachusetts Institute of Technology, co-founder of MegaDetector.
But that promises to change swiftly.
In this month's Technology Feature, journalist @virginiagewin.bsky.social explores how scientists working in the ecology, biodiversity and conservation fields are adopting AI tools to accelerate their research.
Read it here:
www.nature.com/articles/s41...
Read it here:
www.nature.com/articles/s41...
AI carves out a niche in ecology and conservation research - Nature Methods
Scientists who study biodiversity are in the rapid adoption phase of AI. They are finding that what AI can — and can’t — do is shifting rapidly.
www.nature.com
May 5, 2025 at 7:15 PM
Currently, roughly 90–95% of AI usage in biodiversity
and conservation research is simply identifying
a species of interest in gobs of data, says
@sarameghanbeery.bsky.social of the Massachusetts Institute of Technology, co-founder of MegaDetector.
But that promises to change swiftly.
and conservation research is simply identifying
a species of interest in gobs of data, says
@sarameghanbeery.bsky.social of the Massachusetts Institute of Technology, co-founder of MegaDetector.
But that promises to change swiftly.
You may have heard of MegaDetector 🐅
What about a MiniDetector? 🪰
In my opinion flatbug is the closest thing we have to a universal insect detector.
Open source and free to use.
Amazing & groundbreaking work lead by @asgersvenning.bsky.social & @qgeissmann.bsky.social
What about a MiniDetector? 🪰
In my opinion flatbug is the closest thing we have to a universal insect detector.
Open source and free to use.
Amazing & groundbreaking work lead by @asgersvenning.bsky.social & @qgeissmann.bsky.social
I am excited to share that my first first-author paper, “A General Method for Detection and Segmentation of Terrestrial Arthropods in Images,” is now available on preprint.
If you are interested take a look at my blog (asgersvenning.com/flat-bug/) or consider reading the pre-print (linked below).
If you are interested take a look at my blog (asgersvenning.com/flat-bug/) or consider reading the pre-print (linked below).
April 15, 2025 at 12:27 PM
You may have heard of MegaDetector 🐅
What about a MiniDetector? 🪰
In my opinion flatbug is the closest thing we have to a universal insect detector.
Open source and free to use.
Amazing & groundbreaking work lead by @asgersvenning.bsky.social & @qgeissmann.bsky.social
What about a MiniDetector? 🪰
In my opinion flatbug is the closest thing we have to a universal insect detector.
Open source and free to use.
Amazing & groundbreaking work lead by @asgersvenning.bsky.social & @qgeissmann.bsky.social