Over the past few decades, we’ve made some positive advancements to our society. However, some of these advancements have also affected our environment in a big way, and we’re all feeling the repercussions of it. It’s not just us. Just last year, the UN reported that about 1 million plant and animal species are on the verge of going extinct, including penguins.
Penguins are not just adorable animals we can coo over. They’re also important to our ecosystem since their excrement gives nutrition to oceans and lands when they pass through them. Without them, the oceans might be in a worse state.
To study penguin populations, researchers first need to accurately count them. Fortunately, Gramener has teamed up with Microsoft and, utilizing Intel AI, have made a solution to count the penguins. Their solution also has the potential to help us better understand and save penguin populations.
We believe that #AI has the power to help researchers identify what is causing decline in the #Penguin population, and are proud to be using @IntelAI technologies for applications of social impact. https://t.co/H6DwtKYBaO@Microsoft_Green @Microsoft @intel @thepenguinlady https://t.co/rppCYIuOtN
— Gramener (@Gramener) January 21, 2020
Count the penguins, save the penguins
Peter T. Fretwell and Philip N. Trathan have created a study of the current situation of the world’s largest Emperor penguin colony, and it’s pretty dire. They found that the colony has had “unprecedented breeding issues” since 2016. If the largest colony of its species is having that kind of problem, then Antarctica’s penguin populations might disappear altogether. Because of their vulnerability to climate change, this sad reality isn’t far off and they might not make it by 2100.
To study this further and prevent it, Gramener makes use of the Penguin Watch Project and partnered with Microsoft AI for Earth. With this, they created an approach that can approximately count the penguins in clusters using their density. This solution could help researchers lessen obstacles when they manually count penguins from camera traps.
Happy #PenguinAwarenessDay!!! Are you aware that #penguins can teach us about past climates & environmental change? No? Well then check out this video of @DrKeltonMcMahon & I in the field. @LSU_CCE @URIGSO @NSF_OPP @penguin_watch @QuarkExpedition https://t.co/HlOi3MuXX6
— Michael Polito (@MJPolito) January 21, 2020
Solving camera drawbacks
As mentioned, camera traps are tricky due to some elements. A penguin closer to the camera appear larger and takes more image space than the ones at the far end. This makes the perspective distorted altogether. There are also times that penguins overlap or hide other penguins from the camera view. The weather also plays a part in this as it affects visibility. Aside from that, there’s also the setting up of different cameras to capture specifics. It also depends on the population of its dense or coarse.
The impact of their efforts will benefit the researchers with less cost, time, and human involvement. It also helps that it can assist in counting penguins that reside in extreme climatic conditions that might endanger humans.
Today, populations of 11 of the 18 penguin species are decreasing, especially those with restricted ranges living in temperate areas close to humans. This is when these AI models help researchers and organizations to do the job for us to know how to save this species.
Photo credit: The feature photo has been taken by Ian Parker. The demonstration graphic has been provided by Intel as part of their press release and is owned by them.
Sources: Intel Newsroom / Gramener / Stephen Leahy (National Geographic) / Peter T. Fretwell and Philip N. Trathan (Cambridge University Press) / Zooniverse