Indigenous rangers trial drones to track rare rock-wallaby

In this article, you'll read about efforts by First Nations rangers in Australia to track the rare and endangered black-footed rock-wallaby.

Previously monitored with cameras that, while effective, came with the challenge of delayed data analysis for researchers, multi-rotor drones paired with thermal cameras are now being explored as an option to help rangers locate these wallabies quickly. This new tracking method will allow rangers to monitor populations more effectively, as well as monitor for threats like feral cats. 

Researchers turn to drones for that big-picture view of the forest canopy

In other drone news, this Mongabay article discusses how drones are helping researchers understand climate change and deforestation much more efficiently, allowing data to be captured over the course of mere days that would otherwise take a year. The high resolution images captured of forest canopies from above can allow researchers to track changes over time; however, researchers also say that drones work best when complementing on-the-ground fieldwork, not replacing it.

This article also discusses how Indigenous communities are being trained with drones to track deforestation, map resources, and monitor for wildfires, enhancing their abilities to protect and manage forest conservation.

The DeepFaune initiative: a collaborative effort towards the automatic identification of the French fauna in camera-trap images

In this paper, you'll learn about how ML continues to evolve as an invaluable tool for camera trap data analysis. The first results of the DeepFaune project, an effort to "aggregate individual datasets of annotated pictures to train species classification models based on convolutional neural networks, an established deeplearning approach," have now been reported.

The report says that DeepFaune consists of "a two-step pipeline built upon the MegaDetector algorithm for detection (discarding empty pictures and cropping the animal) and a classification model for 18 species or higher-level taxa as well as people and vehicles. The classification model achieved 92% validation accuracy and showed > 90% sensitivity and specificity for many classes."