African elephants surveyed using AI

Satellite monitoring is an unobtrusive technique requiring no ground presence thus eliminating the risk of disturbing species, or of concern for human safety during data collection.

African elephants surveyed using AI. | Picture by special arrangement.

(Subscribe to our Today's Cache newsletter for a quick snapshot of top 5 tech stories. Click here to subscribe for free.)

A new approach for surveying African elephants using satellite imagery and artificial intelligence (AI) could help in solving some of the present challenges in the conservation of the species.

The survey method was developed by an international team of researchers led by University of Oxford. The team detailed their work in a paper titled ‘Using very high-resolution satellite imagery and deep learning to detect and count African elephants in heterogeneous landscapes’, published in the academic journal Remote Sensing in Ecology and Conservation.

A satellite orbiting in space is capable of capturing more than 5,000 square kilometre of imagery in a single run, within minutes. It not only eliminates the risk of double counting, but also makes it possible to conduct repeat surveys at short intervals, an Oxford release explained.

The captured imagery is processed through a deep learning model to detect elephants. The team used a customised dataset of over 1000 elephants in South Africa, to train the model.

“Machines are less prone to error, false negatives and false positives in deep learning algorithms, and are consistent and can be rectified by systematically improving models,” the release said.

 

According to the team, elephants can be detected in satellite imagery with accuracy as high as human detection capabilities. In addition, the model was able to spot elephants in places away from the training data location, and also identify calves even though it was only trained on adults.

The current techniques used for surveying the elephants like using manned aircraft for aerial counts can be costly, hindered by poor visibility, and logistically challenging. Also, erroneous counting could lead to misallocation of scarce conservation resources and misunderstanding population trends, the release noted.

“Satellite monitoring is an unobtrusive technique requiring no ground presence thus eliminating the risk of disturbing species, or of concern for human safety during data collection,” it added.

Further, the remote detection method makes it possible to monitor cross-border areas without terrestrial permits. These areas are usually important to conservation planning.