The Great Wall of China is the only landmark visible from space is a commonly perpetuated myth. However, with modern satellites we can take HD images of far smaller objects like elephants. And with the help of deep learning those objects can be easily detected and tracked.
A team of researchers at the University of Oxford and the University of Bath did just that. Using imagery from Maxar satellites they were to develop a new method for monitoring vulnerable and endangered animals.
An example elephants detected by the CNN model (green boxes), and the true detection labels (red boxes).
The African elephants are of particular interest due to their plummeting population from poaching, retaliatory killing, and habitat fragmentation. To protect them requires an accurate account of their location and numbers. Their size also makes them an ideal candidate for the developing technology.
The Wildlife Conservation Research Unit and Machine Learning Research Group used Maxar’s WorldView-3 satellite imagery and deep learning, TensorFlow and Google Brain, to detect elephants from space with comparable accuracy to human detection capabilities.
The deep learning algorithms are capable of processing massive amounts of data within hours. A task that would take months if done manually. The results produced are consistent and less prone to error, false negatives and false positives.
The researchers believe the results demonstrate the power of modern technology to help serve conservation purposes and the declining global biodiversity.
The team’s open access research publication can be found here.
Does machine learning and its applications across industries interested you? Whether you just want to pick up a few new skills or launch a new career, Lantern Institute provides various programs to help you get started.