Snapshot Serengeti uses AI to aid conservation efforts

Conservation of vulnerable and endangered species is an ongoing, critical venture. Many species face dangers, both natural and man made, which threaten their continued existence. Conservationists have long been fighting to prevent extinction, but it’s an increasingly difficult battle. Snapshot Serengeti is an innovation in artificial intelligence which is helping to track and identify animals with machine learning.

Tanzania is home to hundreds of endangered species, many of which are in danger from natural predators and poachers. Over vast open spaces, conservation relies on up-to-date information to track and protect vulnerable animals.

Unfortunately, it isn’t feasible to track these animals in person. Some are wary of human contact and stay hidden, while others are dangerous predators themselves. Beyond this, there simply aren’t enough conservationists to carefully and reliably track entire populations around the Serengeti without causing considerable disturbance. Surveillance technology is an invaluable asset for monitoring nocturnal travel and capturing footage of any passers-by.

The biggest issue with deploying motion-sensor cameras is that it takes an incredibly long time for volunteers to go through the pictures and correctly identify any animals they see. Academic researchers recently posted that artificial intelligence could carry out this effort much faster without a drop in accuracy.

The team of researchers, led by Prof. Jeff Clune, used a neural network which allows a computer to mimic thought. The network was taught how to identify 48 species correctly, as well as the number of each animal present and what they are doing.

This was done by feeding the neural network over 3 million pictures labeled by human volunteers over many years, courtesy of Snapshot Serengeti. The AI learned from these pictures and was able to apply the data to 99.3 percent of images with no dip in accuracy.

snapshot serengeti

What this means is that future conservation efforts can make greater use of motion-sensors with greater speed and accuracy. At present, cameras only need to be checked to replace their batteries every couple of months, apart from instances of animals damaging the cameras.

With a neural network checking the images and monitoring animal populations, this speeds up the data integration, making conservation efforts more effective.

Snapshot Serengeti is an ongoing effort to aid conservation of animals throughout the Serengeti National Park. While AI and machine learning may one day allow for instantaneous live monitoring of endangered species, the project could still use help from volunteers. By visiting Snapshot Serengeti, anyone can aid the effort.

By browsing through beautiful pictures of the Serengeti and identifying any animals in the pictures, volunteers worldwide can contribute to the process helping to save so many precious animals.

The site also features a tutorial and helpful information for anyone who wants to help but doesn’t know a water buffalo from an Impala or isn’t quite up to date on their big cats. Computers are certainly a huge help, but it will still take the efforts of a lot of people working together to protect the many endangered species of the Serengeti.

Ronan Daly

Ronan Daly is a writer who specialises in Technology and Science. With a Masters Degree in English, and over a decade's academic experience, Ronan has brought a breezy, learned style to My Good Planet, making occasionally complex material accessible and understandable to all.