How the Bangor University in Wales uses Hasty to analyze climate change's effect on maritime eco systems.
About a quarter of our users in Hasty are part of the research-community. We reached out to them to learn more about their research. Some of the stories were so great that we asked the researchers to write a summary that we can share on our blog.
Eleanor Falch from Bangor University in Wales will kick off the series. She’s part of the School of Ocean Sciences. She and her team use visionAI to analyse seabird behaviour, how they’re affected by climate change, and what this means for marine ecosystems in general.
If you want to learn more about the project, reach out to Eleanor @ [email protected]
The guillemot (Latin name: Uria aalge) is a charismatic seabird that is widespread throughout the North Atlantic. These birds are apex predators, meaning that they play an essential role in maintaining the order of many marine food webs and keeping the populations of their prey — such as fish and crustaceans — healthy by picking out old or unwell individuals.
They spend most of their lives at sea but come to rocky cliffs to breed, making them easier to monitor than most marine life, and therefore, they’re a good way of telling how marine ecosystems, which they play a big part in, are doing. Furthermore, like many seabird species, the guillemot is on the UK list for birds of conservation concern, so it is crucial to improve our understanding of how we can help protect them from increasing pressures.
One of the many impacts of climate change is the alteration of weather patterns. How does this affect animals such as guillemots? Previous studies have shown that weather has a substantial impact on animals’ foraging behaviours and even affects the amount of precious energy they use to find food and feed their young. By studying how extreme weather events affect guillemots’ foraging behaviour, we can begin to understand how climate change may affect wildlife and how we can help protect it.
Here at Bangor University, we are studying guillemots that come to breed on Middle Mouse, a small island off the North Coast of Anglesey, North Wales. To do this, we will attach GPS data loggers to a small number of guillemots for several weeks to understand if their foraging movements and behaviours change in response to different environmental factors, including tidal processes, rainfall, wind, and temperature. While the birds are being tracked, we hope to simultaneously conduct at-sea prey surveys in the areas surrounding Middle Mouse Island, onboard Bangor University’s research vessel, RV Prince Madog. This will help support the information gathered using the GPS tags, as we will have a better idea of the abundance and types of prey available in the areas the birds visited at the time they were tracked.
Thanks to funding from the Biodiversity and Ecosystem Evidence and Research Needs Programme, which is funded by the Welsh Government, time-lapse cameras looking out onto the nesting site at Middle Mouse Island are also being used as part of this project. They collect information about when guillemots leave the site to go foraging at sea, supporting the data collected by the GPS loggers and on prey surveys. Analysing the footage from these cameras involves counting each individual bird in a data set of >3000 photographs of this crowded bird colony from last year only. If the birds are to be counted manually in each image for each year of deployment, this would require 100s of hours of researcher effort!
We are currently using Hasty.ai to annotate time-lapse footage taken at the guillemot colony on Middle Mouse during last year’s breeding season. After manually annotating a few images, we unlocked the instance segmentation tool, which suggests objects to annotate as guillemots. We accept correct suggestions and reject the mistakes with a simple click, training the tool to identify these birds better. The Instance Segmentation tool is already very good at identifying guillemots. We are very impressed with its ability to deal with the time-lapse images’ crowded nature (guillemots are highly sociable birds and like to stay very close together!)
Footage from the seabird colony at Middle Mouse Island being annotated using Hasty.ai. Labelled birds are highlighted in blue (guillemots) and purple (razorbills) and annotation suggestions made by the Instance Segmentation tool are surrounded by yellow dotted lines.
The guillemots will be returning to Middle Mouse Island to breed this month, so we will be deploying the time-lapse cameras once again, and we’ll deploy the GPS-tags and carry out prey surveys in June. We will get the data back from these sources in July and use this to find out how guillemot behaviour changed over time and the factors that could be influencing this. In the meantime, we will continue to use Hasty to annotate time-lapse images from last year’s breeding season to develop a model that can be used to count guillemots in this year’s footage, saving us lots of time! A model like this will also be incredibly useful to future studies, helping us understand and protect these incredible animals.
Only 13% of vision AI projects make it to production. With Hasty, we boost that number to 100%.
Our comprehensive vision AI platform is the only one you need to go from raw data to a production-ready model. We can help you with:
All the data and models you create always belong to you and can be exported and used outside of Hasty at any given time entirely for free.
You can try Hasty by signing up for free here. If you are looking for additional services like help with ML engineering, we also offer that. Check out our service offerings here to learn more about how we can help.
It’s time for a non-linear, iterative approach that builds highly performant vision AI applications.
Hasty is now launching Active Learning for labeling. With it, our algorithms review all possible images you can label …