Automatic equine activity detection by convolutional neural networks using accelerometer data

Abstract

In recent years, with a widespread of sensors embedded in all kind of mobile devices, human activity analysis is occurring more often in several domains like healthcare monitoring and fitness tracking. This trend did also enter the equestrian world because monitoring behaviours can yield important information about the health and welfare of horses. In this research, a deep learning-based approach for activity detection of equines is proposed to classify seven activities based on accelerometer data. We propose using Convolutional Neural Networks (CNN) by which features are extracted automatically by using strong computing capabilities. Furthermore, we investigate the impact of the sampling frequency, the time series length and the type of underground on which the data is gathered on the recognition accuracy and evaluate the model on three types of experimental datasets that are compiled of labelled accelerometer data gathered from six different subjects performing seven different activities. Afterwards, a horse-wise cross validation is carried out to investigate the impact of the subjects themselves on the model recognition accuracy. Finally, a slightly adjusted model is validated on different amounts of 50 Hz sensor data. A 99% accuracy can be reached for detecting seven behaviours of a seen horse when the sampling rate is 25 Hz and the time interval is 2.1 s. Four behaviours of an unseen horse can be detected with the same accuracy when the sampling rate is 69 Hz and the time interval is 2.4 s. Moreover, the accuracy of the model for the three datasets decreased on average with about 4.75% when the sampling rate was decreased from 200 Hz to 25 Hz and with 5.27% when the time interval was decreased from 3 s to 0.6 s. In addition, the classification performance of the activity "walk" was not influenced by the type of underground the horse was performing this movement on and even the model could conclude from which underground the data was gathered for three out of four undergrounds with accuracies above 93% at time intervals higher than 1.2 s. This ensures the evaluation of activity patterns in real world circumstances. The performance and ability of the model to generalise is validated on 50 Hz data from different horse types, using tenfold cross validation, reaching a mean classification accuracy of 97.84% and 96.10% when validated on a lame horse and pony, respectively. Moreover, in this work we show that using data from one sensors is at the cost of only 0.24% reduction in accuracy (99.42% vs 99.66%).

Activity Time Budgets—A Potential Tool to Monitor Equine Welfare?

“…This may help to reveal behavioral patterns over more extended periods and determine the influence of environmental conditions on time budgets. Currently, most biotelemetry systems are not yet able to differentiate reliably between different gaits (walk, trot, canter) and between walking and static movements (e.g., stamping, twitching to ward off pests) and thus may determine erroneous gait patterns or too high movement values [ 32 , 47 ]. However, automated tracking methods have been validated in other species (seals, goats, pigs, birds) [ 30 , 31 , 48 , 49 , 50 ] and show great promise for application in equine studies [ 28 , 44 ].…”

“…The time budgets of several welfare relevant behaviors, such as foraging, resting, and lying, can already be accurately determined with commercially available sensors and can be used as welfare indicators to identify welfare problems and monitor the success of interventions [ 32 , 44 ]. Furthermore, real-time analysis of equine behavior may also facilitate early detection of health problems, such as colic, lameness or other painful conditions and accelerate therapeutic interventions [ 4 , 40 , 51 , 52 , 53 , 54 , 55 ].…”

“…Traditionally, this has been done by direct observation which was prone to observer bias, often limited to daylight hours and too time and resource-intensive to be feasible for welfare assessment. However, recent advances in biotelemetry and artificial intelligence (AI) based on sensor (mostly accelerometer) or video analysis provide increased objectivity and enable remote recording of behavioral data, longer observation periods, and quantification of behaviors with higher accuracy and temporal resolution than the human eye allows [ 30 , 31 , 32 ]. Hence, time budget analysis, as an objective, quantitative measure of behavior, now has the potential to become a useful, reliable tool for on-farm assessment of equine welfare and comparison of welfare under different environmental conditions.…”

Feasibility of Wireless Horse Monitoring Using a Kinetic Energy Harvester Model

“…To provide better healthcare for the millions of horses in this industry, anomalies, e.g., colic's, lameness, etc., in the behavior should be detected as early as possible to enable effective treatment and, thereby reducing the risk for possible expensive surgeries. Many of these symptoms can be detected for example by processing continuously collected accelerometer data using heuristics or machine learning models [1,2]. To this end, a wireless sensor node will be attached at the hoof.…”

“…In addition to detecting, e.g., lameness, this information can also be used to calculate per-horse nutrition plans, matching the energy quantity and feeding times to the horse energy consumption [7]. These use cases are described in more detail in, e.g., [1] and motivate the presence of accelerometers in horse wearables. Sensor nodes, in particular for (animal) health applications, are typically limited by physical constraints on the dimensions and therefore the battery capacity is also limited.…”