Deep learning-based animal activity recognition with wearable sensors: Overview, challenges, and future directions

Mao, Axiu; Huang, Endai; Wang, Xiaoshuai; Liu, Kai

doi:10.1016/j.compag.2023.108043

Cited by 25 publications

(14 citation statements)

References 95 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For data processing of wearable sensors, deep learning as a branch of machine learning is able to avoid manually calculating features [52]. However, the aggressive behavior of chickens is small-targeted and multi-scale.…”

Section: Discussionmentioning

confidence: 99%

A Recognition Method for Aggressive Chicken Behavior Based on Machine Learning

Li,

Wang,

Hou

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Aggressive behavior is an important indicator of chicken welfare assessment. At present, the aggressive behavior of chickens typically requires human observation for welfare assessment, and the assessment results are influenced by the subjective judgment of humans. This paper proposes an aggressive chicken behavior identification method based on a hybrid strategy improved Sparrow Search Algorithm combined with Support Vector Machine (ISSA-SVM). Nine-axis inertial sensors were used to collect the behavioral data of chickens. A total of 231-dimensional feature data in the time and frequency domains of the behavioral data were extracted through a 1 s sliding window. To reduce feature redundancy, the initial population is initialized using circle chaotic mapping instead of random initialization of the original sparrow algorithm to increase the uniformity of the initial population distribution in the feature space; adaptive weights are introduced to increase the search range of the early iteration, and the global optimal solution of the previous generation is introduced to improve the global search capability of the algorithm; the optimal solution is perturbed using the dimension-by-dimension mutation strategy of adaptive t-distribution to increase the diversity of the feature distribution. ISSA-SVM reduced the feature dimensionality from 231 to 17, indicating a reduction of 92.6%. The recognition overall accuracy of ISSA-SVM for aggressive chicken behavior was 94.27%, which improved by 1.33% compared to SVM. The results of the experiment show that of all the aggressive chicken behaviors during the 5-day experiment, fighting behavior occurred most frequently from 11: 00 to 13: 00 and from 17: 00 to 18: 00. This study provides a method for the automatic identification of aggressive chicken behavior and can serve as an informational tool for poultry welfare assessment.INDEX TERMS Aggressive behavior, chicken, ISSA-SVM, feature selection, inertial sensors.

show abstract

Section: Discussionmentioning

confidence: 99%

A Recognition Method for Aggressive Chicken Behavior Based on Machine Learning

Li,

Wang,

Hou

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Research on pet behavior prediction has mostly focused on multisensor data rather than video data [4]. This approach focuses on analyzing and predicting pets' movements or activity patterns based on data collected from sensors.…”

Section: Behavior Predictionmentioning

confidence: 99%

“…Understanding and predicting the behavior of pets is crucial for animal welfare, behavior modification, disease prevention, and early detection of diseases [2,3]. Because it relies on manual observation by pet owners or experts, it is time-consuming; there is ongoing research to automate and quantify pet behavior [4]. However, data on pet behavior are often complex, heterogeneous, and frequently incomplete, making its effective analysis and usage challenging.…”

Section: Introductionmentioning

confidence: 99%

Enhanced Pet Behavior Prediction via S2GAN-Based Heterogeneous Data Synthesis

Kim,

Moon

2024

Applied Sciences

View full text Add to dashboard Cite

Heterogeneous data have been used to enhance behavior prediction performance; however, it involves issues such as missing data, which need to be addressed. This paper proposes enhanced pet behavior prediction via Sensor to Skeleton Generative Adversarial Networks (S2GAN)-based heterogeneous data synthesis. The S2GAN model synthesizes the key features of video skeletons based on collected nine-axis sensor data and replaces missing data, thereby enhancing the accuracy of behavior prediction. In this study, data collected from 10 pets in a real-life-like environment were used to conduct recognition experiments on 9 commonly occurring types of indoor behavior. Experimental results confirmed that the proposed S2GAN-based synthesis method effectively resolves possible missing data issues in real environments and significantly improves the performance of the pet behavior prediction model. Additionally, by utilizing data collected under conditions similar to the real environment, the method enables more accurate and reliable behavior prediction. This research demonstrates the importance and utility of synthesizing heterogeneous data in behavior prediction, laying the groundwork for applications in various fields such as abnormal behavior detection and monitoring.

show abstract

“…In activity recognition using wearable devices, domain specificity is highly influenced due to sensor variability [26,35]. Furthermore, AAR faces significant challenges related to domain shift problems, primarily caused by individual variability arising from factors such as size, species, and measurement environments [8,16,19]. In addition, label- Then, unlabeled target data was used to predict animal activities using the model trained with source data (source only).…”

Section: Unsupervised Domain Adaptationmentioning

confidence: 99%

“…Wearable devices, comprising accelerometers, gyroscopes, magnetometers, pressure sensors, and global navigation satellite systems (GNSSs), have garnered increasing popularity in animal monitoring applications. Wearables offer numerous advantages, such as their lightweight nature, compact size, low power consumption, and ease of integration, making them highly suitable for recognizing animal behaviors [7,8]. In this regard, wearable sensors find applications in various contexts, including animal health and welfare [6,9], as well as smart animal farming [10,11] for monitoring animal behavior.…”

Section: Introductionmentioning

confidence: 99%

Unsupervised Domain Adaptation for Mitigating Sensor Variability and Interspecies Heterogeneity in Animal Activity Recognition

Ahn,

Kim,

Jeong

2023

Animals

View full text Add to dashboard Cite

Animal activity recognition (AAR) using wearable sensor data has gained significant attention due to its applications in monitoring and understanding animal behavior. However, two major challenges hinder the development of robust AAR models: domain variability and the difficulty of obtaining labeled datasets. To address this issue, this study intensively investigates the impact of unsupervised domain adaptation (UDA) for AAR. We compared three distinct types of UDA techniques: minimizing divergence-based, adversarial-based, and reconstruction-based approaches. By leveraging UDA, AAR classifiers enable the model to learn domain-invariant features, allowing classifiers trained on the source domain to perform well on the target domain without labels. We evaluated the effectiveness of UDA techniques using dog movement sensor data and additional data from horses. The application of UDA across sensor positions (neck and back), sizes (middle-sized and large-sized), and gender (female and male) within the dog data, as well as across species (dog and horses), exhibits significant improvements in the classification performance and reduced the domain discrepancy. The results highlight the potential of UDA to mitigate the domain shift and enhance AAR in various settings and for different animal species, providing valuable insights for practical applications in real-world scenarios where labeled data is scarce.

show abstract

Deep learning-based animal activity recognition with wearable sensors: Overview, challenges, and future directions

Cited by 25 publications

References 95 publications

A Recognition Method for Aggressive Chicken Behavior Based on Machine Learning

A Recognition Method for Aggressive Chicken Behavior Based on Machine Learning

Enhanced Pet Behavior Prediction via S2GAN-Based Heterogeneous Data Synthesis

Unsupervised Domain Adaptation for Mitigating Sensor Variability and Interspecies Heterogeneity in Animal Activity Recognition

Contact Info

Product

Resources

About