A new iPad app is asking users to “tag” tigers, to help researchers with counting and tracking the animals.Images gathered from across the internet are fed into the app, which is set up as a game.Players score points by tagging the images that contain tigers, as well as answering questions about the big cats’ surroundings and what they are doing.Eventually the team aims to combine the data with facial recognition software, to help track individual animals.The project is called Wildsense and it emerges from research at the University of Surrey, where a team of computer scientists has a particular passion for wildlife conservation – and they want to share that with the public.They have developed the app as a tool to turn the internet’s vast wealth of images, and the help of willing iPad users, into valuable data for conservationists.Aaron Mason, a PhD student with a background in software and app development, told the BBC: “It pulls in photos from the internet which have the word tiger associated with them – but that brings up all manner of images, from baseball teams, to cuddly toys, to furry cats and pets. We want to get rid of that noise.” Hidden among that noise will be photos of real tigers in the wild, many of them uploaded by tourists to public sites like Flickr: a largely untapped resource for conservationists monitoring the species. So first of all, users will be asked: Is there a real tiger in this photo?After that, they will pinpoint its face in the image, and give some more, simple details about the picture.”People can interact with the photos and provide more context,” Mr Mason said.This will include the type of surroundings, the weather, and what the beast in the photo is up to – is it running, sleeping or eating, for example.All of this data will be put together with whatever information can be gleaned automatically from the photo, such as its location. That might be from the original camera’s GPS tag, or simple keywords like “India”. Then comes the job of analysing the results, which falls to Mr Mason and his colleagues. They are hopeful that the app – and anyone who plays the game – will make a useful contribution to monitoring the numbers, distribution and activity of tigers in the wild.There is certainly no shortage of the raw resource, because the internet is famously full of pictures.”There’s so many photos online, we still haven’t even processed them all,” Mr Mason said.”We’re hoping after the experiment we’ll understand better where they come from – is it guides, is it tourists, or is it something else?”Once the photos have been annotated, the team hopes to apply facial recognition software to the tagged tigers. Trials have shown some promise at distinguishing individuals in this way.And the team had some previous success identifying specific tigers from their stripes, in an ongoing project called Tiger Nation. In that case, people who have seen wild tigers in India upload their own snapshots as part of a more active effort to help monitor the population.It proved popular; Tiger Nation currently has several thousand active users, Mr Mason said. That is about as many participants as there are tigers alive in the wild.By contrast, the Wildsense app enlists anyone with an iPad to help, by processing any image the researchers can lay their hands on. They are already keen to add photos that haven’t simply been scraped from the web.”There are other sources which we are beginning to access,” said Prof Paul Krause, who runs the Surrey laboratory where Wildsense was developed. “For example, a lot of professional photographers publish their photographs in libraries. We need permission to access those photos – but we’re working with two or three people to grant us that access.”They’re richer in information, because typically they will include time and date information, and we get a high quality image.”The team also has experience gathering its own photos, with the help of different specialised technology.”For this experiment we’re mainly using photos from the internet, uploaded by tourists. But in the past we’ve done experiments where we deployed camera traps in the field – in the jungle in India for example – which sent us photos directly,” Mr Mason explained.So by combining these image sources – and enlisting the help of citizen scientists and facial recognition algorithms – there is the potential to track the movements of individual tigers in the wild, in an entirely non-invasive way.”That’s our aim,” Mr Mason said.
May 29, 2021
May 28, 2021