Using AI to Monitor Trackside Britain’s Wildlife

We’re working at the cutting edge of conservation – using artificial intelligence to rapidly process vast amounts of field data to protect British wildlife.  

Britain is home to an array of wildlife – from blackbirds in trees to hedgehogs scurrying through the undergrowth beneath them.  

Yet these animals face a growing series of challenges. The UK is already one of the most nature-depleted countries in the world, and ongoing threats such as climate change and habitat loss will only further add to this biodiversity problem.  

Railway tracks may not be the first place that come to mind when we think of spaces for nature, but trainlines run through major wildlife habitats across Britain, home to many rare and protected species. 

We’ve been working with Network Rail and Google Cloud at the cutting-edge of wildlife monitoring along these railway tracks. Network Rail owns more than 52,000 hectares of land, and many of these areas play a key role in protecting biodiversity. Building upon our work using technology to monitor wildlife in Cameroon, we’re developing ways to rapidly identify the birds and mammals living in these trackside habitats, providing essential information and guidance for Network Rail ecologists working to protect nature at these sites.  

How can artificial intelligence help protect wildlife?

When working to protect wildlife, it’s essential that we understand where species are living, how many of them are in a certain area, and how well their populations are doing. Camera traps and acoustic monitoring allow researchers to easily and remotely gather raw data that would otherwise take days or weeks of their time – but all this raw data needs to be analysed before it can be used to feed into management guidance and conservation action.  

Analysing tens of thousands of images is a laborious process and carrying it out manually slows down the process of turning conservationists’ research into action. Our Instant Wild platform – where citizen scientists can help identify animals from photos and video footage – helps speed up the process and welcomes members of the public to be conservationists from the comfort of their armchairs, but exciting new technologies also hold the key to speedier wildlife monitoring.

Machine learning – a common form of artificial intelligence where computer algorithms are trained to recognise particular patterns – allows us to quickly identify species through automatic recognition. For example, an algorithm trained to identify bird calls can rapidly process thousands of hours of data to identify which species are present within a monitored habitat. Huge amounts of data can be processed, and wildlife trends can be tracked in almost real time, freeing up conservationists to take the action needed to protect these species.  

Monitoring London Trackside Wildlife

As part of our work with Network Rail and Google Cloud, over spring and summer 2022 we collected 3000 hours of audio and 40,000 images from acoustic monitors and camera traps placed at our three pilot sites across London. At this large a scale, processing all the data by human hand alone would have been a formidable and time-consuming task. However, through working with partners at Google Cloud to use machine learning, the animals living alongside the tracks could be rapidly identified.  

Six bat species and over 30 bird species were identified – including Eurasian blackcaps, blackbirds and great tits – alongside foxes, deer and hedgehogs, highlighting just how many species can be found using the green spaces alongside railway tracks. More importantly, the work achieved its primary aim - showing that the technology works and can be used to effectively survey wildlife at scale. 

Using AI-led Technology for Conservation

Now that the team have shown proof of concept – using machine learning to analyse huge amounts of wildlife monitoring data – they can now look at expanding this to other areas to support Network Rail’s understanding of wildlife along their tracks.  For example, bats are likely using railway bridges for roosting, so using monitors to identify exactly where they’re roosting can be used to help protect them.

Showing that AI can be used effectively to monitor wildlife is ground-breaking for conservation, as it opens the door for scientists and conservationists to work smarter and answer what were previously impossible questions.  With this knowledge, we can gain further understanding of the threats and challenges animals face and act faster to protect them.  

One of the biggest threats we currently face is climate change, and as species move to new ranges in response to changing conditions, monitoring at scale will help conservationists understand where they’re moving to and where to take protective action. Using and developing new and revolutionary conservation technologies is an integral part of our work to restore the natural world.