Behaviour Classification on Giraffes (Giraffa camelopardalis) Using Machine Learning Algorithms on Triaxial Acceleration Data of Two Commonly Used GPS Devices and Its Possible Application for Their Management and Conservation

Abstract

Averting today’s loss of biodiversity and ecosystem services can be achieved through conservation efforts, especially of keystone species. Giraffes (Giraffa camelopardalis) play an important role in sustaining Africa’s ecosystems, but are ‘vulnerable’ according to the IUCN Red List since 2016. Monitoring an animal’s behavior in the wild helps to develop and assess their conservation management. One mechanism for remote tracking of wildlife behavior is to attach accelerometers to animals to record their body movement. We tested two different commercially available high-resolution accelerometers, e-obs and Africa Wildlife Tracking (AWT), attached to the top of the heads of three captive giraffes and analyzed the accuracy of automatic behavior classifications, focused on the Random Forests algorithm. For both accelerometers, behaviors of lower variety in head and neck movements could be better predicted (i.e., feeding above eye level, mean prediction accuracy e-obs/AWT: 97.6%/99.7%; drinking: 96.7%/97.0%) than those with a higher variety of body postures (such as standing: 90.7–91.0%/75.2–76.7%; rumination: 89.6–91.6%/53.5–86.5%). Nonetheless both devices come with limitations and especially the AWT needs technological adaptations before applying it on animals in the wild. Nevertheless, looking at the prediction results, both are promising accelerometers for behavioral classification of giraffes. Therefore, these devices when applied to free-ranging animals, in combination with GPS tracking, can contribute greatly to the conservation of giraffes.

Wildlife Monitoring on the Edge: A Performance Evaluation of Embedded Neural Networks on Microcontrollers for Animal Behavior Classification

“…In previous state-of-the-art solutions, such as [ 9 , 10 , 11 , 12 , 13 , 14 ], authors evaluated the classifier in a local computer without providing on-device or cloud-based real-time animal behavior classification. This fact restricts their implementation in a real scenario, since predicting behavioral patterns of the animals after removing the monitoring device from them and processing the stored information is not the optimal use case.…”

“…Our proposed system is capable of performing real-time online classification, allowing expert biologists to access the data as soon as it is predicted by the collar devices. In terms of battery life and power consumption, the aforementioned studies report a battery life of 12–14 days [ 9 ], 14 days [ 10 ], 7 days [ 12 ], and 17 days [ 13 ], respectively. When using the lower power consumption microcontroller from the two options that were considered in this work, we obtain an estimated battery life of 88 days in the worst case, which is 5 times more battery life than [ 13 ], taking into consideration that, in our work, we also include the processing of the data and the prediction from the ENN in this estimation.…”

“…When using the lower power consumption microcontroller from the two options that were considered in this work, we obtain an estimated battery life of 88 days in the worst case, which is 5 times more battery life than [ 13 ], taking into consideration that, in our work, we also include the processing of the data and the prediction from the ENN in this estimation. However, if a midpoint case is considered, the estimated battery life is more than 15 times higher than the one from [ 13 ].…”

“…This kind of network normally uses two types of wireless connections: one for short range, such as Bluetooth or ZigBee, and another one for large range or Internet connection, such as WiFi, GPRS, or LoRA [ 8 ]. Recent studies [ 9 , 10 , 11 , 12 , 13 , 14 ] have proposed collar devices consisting of a GPS module and an IMU (or an accelerometer only) for animal behavior recognition. In [ 9 ], the authors present a device which collects information from a 3-axis accelerometer and a GPS.…”

“…The system classifies between four different behaviors (walking, feeding, lying, and standing) using a Random Forest algorithm with an 94.3% accuracy in an offline manner. In [ 13 ], the authors compared two commercial devices that provide tri-axial acceleration data and GPS information. These devices are used to obtain data related to different behaviors from giraffes, which are then used to train and evaluate a machine learning algorithm (Random Forest), achieving a mean accuracy of 88.4% over six different behaviors when using cross validation.…”