Classification of broiler behaviors using triaxial accelerometer and machine learning

Authors: YANG, XIAO - University Of Tennessee; ZHAO, YANG - University Of Tennessee; STREET, GARRETT - Mississippi State University; Huang, Yanbo; TO, S.D. FILIP - Mississippi State University; Purswell, Joseph - Jody

Submitted to: Animal-The International Journal of Animal Biosciences
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/27/2021
Publication Date: 6/5/2021
Citation: Yang, X., Zhao, Y., Street, G.M., Huang, Y., To, S., Purswell, J.L. 2021. Classification of broiler behaviors using triaxial accelerometer and machine learning. Animal-The International Journal of Animal Biosciences. https://doi.org/10.1016/j.animal.2021.100269.
DOI: https://doi.org/10.1016/j.animal.2021.100269

Interpretive Summary: Understanding broiler behaviors is important for poultry farm management. Scientists from University of Tennessee, Mississippi State University, USDA-ARS (Crop Production Systems Research Unit at Stoneville, MS and Poultry Research Unit at Mississippi State, MS) have collaboratively investigated the classification of broiler behaviors by analyzing data from wearable accelerometers using two machine learning models. The results indicated that the broiler behaviors could be classified well and the performances of different machine learning models varied.

Technical Abstract: Understanding broiler behaviors provides important implications for animal well-being and farm management. The objectives of this study were to classify specific broiler behaviors by analyzing data from wearable accelerometers using two machine learning models, K-Nearest Neighbor (KNN) and Support Vector Machine (SVM). Lightweight triaxial accelerometers were used to record accelerations of nine 7-week-old broilers at a sampling frequency of 40Hz. A total of 261.6-min data were labeled for four behaviors – walking, resting, feeding and drinking. Instantaneous motion features including magnitude area, vector magnitude, movement variation, energy, and entropy were extracted and stored in a dataset which was then segmented by one of the six window lengths (1, 3, 5, 7, 10 and 20 s) with 50% overlap between consecutive windows. The mean, variation, standard deviation, minimum and maximum of each instantaneous motion feature and two-way correlations of acceleration data were calculated within each window, yielding total 43 statistic features for training and testing of machine learning models. Performance of the models were evaluated using pure behavior datasets (single behavior type per dataset) and continuous behavior datasets (continuous recording that involved multiple behavior types per dataset). For pure behavior datasets, both KNN and SVM models showed high sensitivities in classifying broiler resting (87% and 85%, respectively) and walking (99% and 99%, respectively). The accuracies of SVM were higher than KNN in differentiating feeding (88% and 75%, respectively) and drinking (83% and 62%, respectively) behaviors. Sliding window with 1-s length yielded the best performance for classifying continuous behavior datasets. The performance of classification model generally improved as more birds were included for training. In conclusion, classification of specific broiler behaviors can be achieved by recording bird triaxial accelerations and analyzing acceleration data through machine learning. Performances of different machine learning models differ in classifying specific broiler behaviors.