This teenager invented an inexpensive tool to detect elephant poachers in real time

Anika Puri, just 17, has designed a machine learning-based model that analyzes the movement patterns of humans and elephants.

When Anika Puri traveled to India with her family, she was surprised to find a Bombay market lined with rows of jewelry and ivory statues. Worldwide, the ivory trade has been illegal for more than 30 years and elephant hunting has been banned in India since the 1970s.

Curious, Puri did some research and found a shocking statistic: Africa’s forest elephant population had declined by around 62% between 2002 and 2011. Years later, the numbers continue to drop. Puri wanted to do something to help protect this species and others that continue to be threatened by poaching.

Drones are currently used to detect and capture images of poachers, and they are not as accurate, the teenager explains. But after watching videos of elephants and humans, he saw how the two differed greatly in how they moved: their speed, turn patterns and other movements.

He realized that this disparity between these two movement patterns could be used to increase the accuracy of detecting potential poachers.

In two years, Puri created ElSa, a low-cost prototype of machine learning-based software that analyzes movement patterns in thermal infrared videos of humans and elephants. Puri claims the software is four times more accurate than the most advanced detection methods. It also eliminates the need for expensive high-resolution thermal cameras, which can cost thousands of dollars, he says. ElSa uses a FLIR ONE Pro thermal imaging camera, around $250, which attaches to a stock iPhone 6. The camera and the iPhone are connected to a drone, and the system counts in real timewhen flying over the parks, whether the objects below are humans or elephants.

Puri submitted his project to this year’s Regeneron International Science and Engineering Fair, the world’s largest international science, technology, engineering and math competition. Its possible impact on society has earned it the Peggy Scripps Prize for Science Communicationand also won first prize in the earth and environmental science category of the competition.



To create his model, Puri first found movement patterns of humans and elephants using the Reference IR Dataset for Surveillance with Aerial Intelligence (BIRDSAI), a dataset collected using a thermal infrared camera attached to an unmanned aerial vehicle (UAV) in several protected areas in Africa. Reviewing the data, Puri identified 516 time series taken from videos that captured moving humans or elephants.

Puri used a machine learning algorithm to train a model that would classify a figure as an elephant or a human based on its speed, group size, turning radius, number of turns and other patterns. He used 372 sets: 300 elephant moves and 72 human moves. The remaining 144 were used to test his model on data he had never seen before. When tested with the BIRDSAI dataset, their model was able to detect humans with over 90% accuracy.

Going through

Pictures: Society for Science

Leave a Comment