Using Artificial Intelligence to catch wildlife poachers in almost real-time

Poaching of wildlife is a big problem globally and especially in Afrika. This can be witnessed in the recently released statistics on Rhino deaths as a result of poaching by South Africa's Department of Environmental Affairs.

However, although the number of rhinos poached is alarmingly high compared to several years ago in South Africa's Kruger National Park, for 2017 the number of rhinos poached decreased from 1,054 to 1,028 thanks to a large part on the anti-poaching technology being used in the Kruger National Park. This number could likely even decrease further if technology such as the one developed by the University of Southern California could be widely used in game parks.

The Kruger National Park uses a surveillance solution known as the Meerkat. It is made up of a suite of radar and electro-optic sensors that will detect, classify, monitor and track humans moving in the Kruger National Park over a wide area.

A screenshot of the field test environment with annotated figures. | University of Southern California

Lately, the University of Southern California has joined others who are developing technology solutions to combat wildlife poaching. The American university developed technology that uses Artificial Intelligence (AI) and game theory to predict where next the poachers are likely to be. Their "Systematic POacher deTector (SPOT)" is also able to differentiate humans from animals.

"[W]e developed a system, SPOT, to automatically detect poachers as well as animals in thermal infrared UAV videos taken at night in near real time, which
shows that modern computer vision techniques are capable of conquering difficulties that have not been addressed before. This system works in varying situations and does not require the users to adjust any parameters when they use it. Thus, it is easily accessible to non-expert users. Furthermore, the system can detect poachers in near real time with either good or bad network connectivity. The system has been tested in the field, and will be deployed in the near future in several national parks in Africa, including one in Botswana. SPOT opens the door for exciting new research questions in object detection in difficult videos, and for new anti-poaching
strategies utilizing UAVs in the field."

SPOT is not the first technology solution to use AI and drones in anti-poaching, but it is likely the one that will probably be most efficient yet. In 2017, A new collaboration project between Neurala and the Lindbergh Foundation’s AirShepherd program was announced, at the time, it was using AI and drones to combat the poaching of elephants and rhinos in Afrika, but not in almost real-time, as SPOT is reported to be capable of doing.

Cover image credit: Herd of elephants at the Pilanesberg National Park in South Africa. |