Browsing by Author "Braz, Fernando"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
- Exploring the Potential of Machine Learning Algorithms Associated with the Use of Inertial Sensors for Goat Kidding DetectionPublication . Gonçalves, Pedro; Marques, Maria do Rosário; Belo, Ana Teresa; Monteiro, António; Morais, João; Riegel, Ivo; Braz, FernandoThe autonomous identification of animal births has a significant added value, since it enables for a prompt timely human intervention in the process, protecting the young and the mothers’ health, without requiring continuous human surveillance. Wearable inertial sensors have been employed for a variety of animal monitoring applications, thanks to their low cost and the fact that they allow less invasive monitoring process. Alarms triggered by the occurrence of events must be generated close to the events to avoid delays caused by communication latency, which is why this type of mechanism is typically implemented at the network’s edge and integrated with existing auxiliary mechanisms on the Internet. Although the detection of births in cattle has been carried out commercially for some years, there is no solution for small ruminants, especially goats, where the literature does not even report any attempts. The current work consisted of a first attempt at developing an automatic birth monitor using inertial sensing, as well as detection techniques based on Machine Learning, implemented in a network edge device to assure real-time alarm triggering. Thus, two concept drift detection techniques and seven kidding detection mechanisms were developed using data classification models. The work also includes the testing and comparison of learning results, both in terms of accuracy and of computational costs of the detection module, for algorithms implemented. The results revealed that, despite their simplicity, concept drift algorithms do not allow kidding detection, whereas classification-algorithm-based static learning models do, despite the unbalanced character of the dataset and its reduced size. The learning findings are quite promising in terms of computational cost and its suitability for deployment on edge devices. The algorithm demonstrates behavior changes four hours before kidding and allows for the identification of the kidding hour with an accuracy of 61%, as well as the capacity to improve the overall learning process with a larger dataset
- Goat Kidding DatasetPublication . Gonçalves, Pedro; Marques, Maria R.; Belo, Ana T.; Monteiro, António; Braz, FernandoThe detection of kidding in production animals is of the utmost importance, given the frequency of problems associated with the process, and the fact that timely human help can be a safeguard for the well-being of the mother and kid. The continuous human monitoring of the process is expensive, given the uncertainty of when it will occur, so the establishment of an autonomous mechanism that does so would allow calling the human responsible who could intervene at the opportune moment. The present dataset consists of data from the sensorization of 16 pregnant and two non-pregnant Charnequeira goats, during a period of four weeks, the kidding period. The data include measurements from neck to floor height, measured by ultrasound and accelerometry data measured by an accelerometer existing at the monitoring collar. Data was continuously sampled throughout the experiment every 10 s. The goats were monitored both in the goat shelter (day and night) and during the grazing period in the pasture. The births of the animals were also registered, both in terms of the time at which they took place, but also with details regarding how they took place and the number of offspring, and notes were also added.
- Sheep Nocturnal Activity DatasetPublication . Monteiro, António; Gonçalves, Pedro; Marques, Maria R.; Belo, Ana T.; Braz, FernandoMonitoring sheep’s behavior is of paramount importance, because deviations from normal patterns may indicate nutritional, thermal or social stress, changes in reproductive status, health issues, or predator attacks. The night period, despite being a more restful period in which animals are theoretically sleeping and resting, represents approximately half of the life cycle of animals; therefore, its study is of immense interest. Wearable sensors have become a widely recognized technique for monitoring activity, both for their precision and the ease with which the sensorized data can be analyzed. The present dataset consists of data from the sensorization of 18 Serra da Estrela sheep, during the nocturnal period between 18 November 2021 and 16 February 2022. The data contain measurements taken by ultrasound and accelerometry of the height from neck to ground, as well as measurements taken by an accelerometer in the monitoring collar. Data were collected every 10 s when the animals were in the shelter. With the collection of data from various sensors, active and inactive periods can be identified throughout the night, quantifying the number and average time of those periods.