Who Says All Koalas look Alike? Not using Drones for Data Collection and AI they Don't

Also being relatively small, koalas are easy to miss; so the accurate data collection is more suspect. This issue has varying importance given the size and the area covered (think elephants or giraffes, which are easier to locate given their larger size, but may be in more remote areas further from where humans reside and therefore less accessible).

Now, drones equipped with thermal imaging can find the koalas – and AI can identify the specific animal, to avoid counting it twice and provide greater insight into behavior and health.

“Effective wildlife management relies on the accurate and precise detection of individual animals. These can be challenging data to collect for many cryptic species, particularly those that live in complex structural environments. This study introduces a new automated method for detection using published object detection algorithms to detect their heat signatures in RPAS-derived thermal imaging. As an initial case study we used this new approach to detect koalas (Phascolarctus cinereus), and validated the approach using ground surveys of tracked radio-collared koalas in Petrie, Queensland. The automated method yielded a higher probability of detection (68–100%), higher precision (43–71%), lower root mean square error (RMSE), and lower mean absolute error (MAE) than manual assessment of the RPAS-derived thermal imagery in a comparable amount of time. This new approach allows for more reliable, less invasive detection of koalas in their natural habitat. This new detection methodology has great potential to inform and improve management decisions for threatened species, and other difficult to survey species.”

The data collection is something that we are used to seeing with drones, but the combination of AI is interesting in this application. Interpreting the data manually is tedious and time-consuming, as well as error prone. A commercial application such as reviewing aerial photography from a drone for a long pipeline inspection is very analogous, or even CCTV camera’s looking for bad guys: looking at large amounts of video data is difficult, boring, and prone to errors.

It seems, however, that koalas have individual heat signatures. That allows scientists to identify individual animals with some confidence.

Again from the piece:

“Automated detection of individual animals in remotely sensed imagery can reduce bias and increase accuracy and precision of wildlife surveys, but few methods have been developed and tested in the field. For mammals, the automated detection methods shown to be most accurate thus far have been applied to thermal imagery, as the large temperature gradient between mammals and their background environment allows computer vision to easily detect and count their thermal signatures.

The monitoring and management of koalas (Phascolarctus cinereus), an Australian mammal species of conservation concern, has the potential to benefit greatly from development of a robust automated detection method using RPAS and thermal imagery^18,19. Koala populations are often widely and unevenly distributed and frequently occur on private property, making them difficult and time-consuming to survey accurately by direct observation^18,20. They are also cryptic in nature and inhabit environments with complex canopy cover, which significantly lowers the probability of detecting all individuals through direct observation both on the ground and in colour photographic imaging.

Candidate ‘koala’ signatures were then detected from the averaged heat map as shown in Fig. 2.”