Recent photos of Australian wildlife in bushfire recovery areas have been captured and analysed using artificial intelligence, shedding light on the journey of vulnerable species recovering from natural disasters. Researchers from WWF and Conservation International, in collaboration with local land managers, collected over 7 million photos from 1,100 sensor-activated cameras placed in eight bushfire-affected parts of Australia.
Collaboration between Organizations to Collect Data
The cameras captured various animals, including wombats, echidnas, dingo pups, and koalas, with some of the species captured in groups for the first time after the bushfires. The Google AI technology, called Wildlife Insights, was trained on 4 million images of over 150 Australian animals and tracked their recovery in the years following the fires. Although the system initially mistook some species, it can now identify kangaroos and wombats with over 90% accuracy.
Improved Accuracy in Identifying Species
The cameras have been in different locations for between a few months and three years, and they have also captured invasive species such as foxes, feral cats, pigs, and cane toads. Dr Emma Spencer, a researcher and coordinator of WWF Australia’s eyes on recovery program, said the discovery of dunnarts on Kangaroo Island, where bushfires devastated up to 90% of their habitat in 2020, was particularly exciting. Images of koalas moving around on the ground indicate that the animals are still recovering from the fires, and Spencer believes that they may be searching for new habitat due to the loss of their previous habitat.
Implications for Future Bushfires
The cameras’ data can help researchers and land managers to identify threatened species quickly, which can aid in their response to future fires. Although the project has highlighted signs of wildlife recovery, the potential for the next bushfire season is not far away, as vegetation growth following heavy rainfall in the past three years could lead to higher bushfire risk. Spencer hopes that the results can inform future fire events, which are expected to increase due to climate change.
