On the use of on-cow accelerometers for the classification of behaviours in dairy barns

Abstract

Analysing behaviours can provide insight into the health and overall well-being of dairy cows. Automatic monitoring systems using e.g., accelerometers are becoming increasingly important to accurately quantify cows' behaviours as the herd size increases. The aim of this study is to automatically classify cows' behaviours by comparing leg- and neck-mounted accelerometers, and to study the effect of the sampling rate and the number of accelerometer axes logged on the classification performances. Lying, standing, and feeding behaviours of 16 different lactating dairy cows were logged for 6h with 3D-accelerometers. The behaviours were simultaneously recorded using visual observation and video recordings as a reference. Different features were extracted from the raw data and machine learning algorithms were used for the classification. The classification models using combined data of the neck- and the leg-mounted accelerometers have classified the three behaviours with high precision (80-99%) and sensitivity (87-99%). For the leg-mounted accelerometer, lying behaviour was classified with high precision (99%) and sensitivity (98%). Feeding was classified more accurately by the neck-mounted versus the leg-mounted accelerometer (precision 92% versus 80%; sensitivity 97% versus 88%). Standing was the most difficult behaviour to classify when only one accelerometer was used. In addition, the classification performances were not highly influenced when only X, X and Z, or Z and Y axes were used for the classification instead of three axes, especially for the neck-mounted accelerometer. Moreover, the accuracy of the models decreased with about 20% when the sampling rate was decreased from 1Hz to 0.05Hz.

Classification of ingestive-related cow behaviours using RumiWatch halter and neck-mounted accelerometers

“…Every one minute time window was assigned a label to refer to feeding, ruminating, and other activity (non-ingestive), respectively, based on the behaviour that was present during the largest proportion of that minute. Instead of removing the small number of samples of drinking behaviour (i.e., less than 2%), they were considered as feeding as per the methodology used by (Benaissa et al, 2017). As 6 hours of visual observation were available for 10 cows (Table 2), 3600 samples of observed behaviours were obtained (i.e., 3600 min).…”

“…In the latter study, the tilt of the accelerometer axes was used for the classification. However, as explained in (Benaissa et al, 2017) , this method is impractical in real situations where a slight movement of the sensor could change the reported tilt of the axes within the same behaviour. For neck-mounted accelerometers, Martiskainen et al, (2009) developed a method for automatically measuring and recognising several behaviours of dairy cows, including feeding and ruminating behaviours, based on a multi-class support vector machine (SVM).…”

Inferring an animal's environment through biologging: quantifying the environmental contribution to animal movement

“…Furthermore, we have found no substantial inter-or intra-cow activity classification performance differences. We thus considered the SVM activity classification models good enough to reliable predict the activity types based on the movement sensor data, even more so because the classification performance was higher or comparable to other cow activity classification studies [4,32,33].…”

“…As animal movement is inherently multifaceted, with aspects related to the movement of the animal through the landscape and aspects related to the movement of body parts, the movement process cannot be described with simpli ed descriptors without loss of information. On the contrary, a plethora of emergent patterns can be identi ed through these multifaceted movement descriptors, e.g., activity types (such as walking, foraging or resting) and collective movement properties [4,5]. Technological advancements in the eld of biologging currently allow for data on animal movement to be acquired at ner temporal and spatial scales and in increasing volumes, e.g., data on animal movement speed, movement path tortuosity, triaxial acceleration of body parts, and heart rate patterns can now relatively easily be acquired [6][7][8].…”

Wireless energy transfer by means of inductive coupling for dairy cow health monitoring

“…Automatic monitoring systems can be implemented for the detection of illnesses, predicting the calving moment, and tracking the movement and location of the animal [8,9,10,11,12]. On-body sensors allow measuring different 30 parameters of the animal, which can be wirelessly transferred to a back-end server for data interpretation [13,14].…”