23 – The technology revolution: improving species detection and monitoring^307carefully in the higher resolution colour images, speeding up significantly the
manual process of going through the footage. The efficiency of this system will be
compared with a higher resolution (640 × 512 pixels) high-end (~A$6000) thermal
sensor, with better detection capabilities but significantly higher cost.
The first aerial detection of an LTK (Fig. 23.1b–c) with the low-resolution
sensor was obtained during preliminary field trials, which also indicated that the
animals are not disturbed by the noise of the rotors. More extensive field trials of
the system are currently planned, following systematic f light patterns. The use of
new (and often expensive) technology has to be justified by an improvement in
monitoring capability and/or efficiency. To this end, visual surveys were also
conducted (daytime and nighttime (by either spotlighting or using a handheld
thermal sensor)). The detectability achieved by the different approaches (including
the drone with both low-resolution and high-end thermal sensors), their rates of
false detections (how often another species is detected as an LTK) and ultimately
their efficiency at monitoring occupancy of the species will be estimated and
compared. If proven reliable, this low-cost system could become a new tool to
monitor LTK and potentially other arboreal mammals (e.g. Bennett’s tree-
kangaroo) without the access limitations associated with walked transects.
Fig. 23.1. (a) Lumholtz’s tree-kangaroos. Photo: J. Hopkinson. (b) Low-resolution thermal image with a LTK
detection. (c) The thermal image in b was taken with the drone passing over a known LTK. Photo: G.
Guillera-Arroita. (d) Technical details of the drone-based thermal detection system: FLIR development kit
non-radiometric thermal sensor controlled with a Raspberry Pi 3 single-board processor, mounted on a
commercial drone (DJI Phantom 3 Advanced); an external GPS unit is used to geotag the pictures.