Deep learning image recognition of cow behavior and an open data set acquired near an automatic milking robot

Abstract

Production animals enjoying good health and well-being are more productive and have a higher output quality. Several technical solutions have been used to monitor the animals’ welfare: those based on computer vision provide cost-efficient and scalable options. In this work, we performed a continuous two-month image acquisition of cows in front of an automatic milking station and divided the data into ten different classes related to the most important activities appearing in the images. The data consisted of almost 19 hours of video, equivalent to more than 1.7 million still images. Based on these imaged, we then implemented a convolutional neural network classifier to recognize the cows' behavior. The network was tested using cross-validation methodology and achieved an 86% precision rate and 85% recall rate in the classification. The data and the Python program code used in this study are made available. An image data set that directly resembles the harsh conditions inside a barn and can be used for deep learning purposes has not been previously made available.

Image recognition method for dangerous behavior of non-stop construction personnel in large airports

“…Reference [7] puts forward a traffic behavior identification method in traffic video based on Kalman filter, which sets the traffic flow and turning events of different roads, tracks the vehicles in traffic video by using YOLO algorithm, and estimates and predicts the speed and position of vehicles by using Kalman filter to complete traffic flow feature identification. Reference [8] puts forward a method of cow image behavior recognition based on deep learning, which divides the data into different categories related to the most important activities in the image, and uses convolutional neural network classifier to identify cow behavior. Reference [9] proposes a symmetric algorithm to capture spatio-temporal information based on behavior changes, uses the improved DeepLabCut key point detection algorithm to locate mouse limbs, and uses ConvLSTM network to extract spatio-temporal information from key point feature map sequence to classify mouse behaviors, thus completing the recognition of mouse movement behaviors.…”

Precision Livestock Farming Research: A Global Scientometric Review

“…13, x 18 of 29 subdivision of machine learning that employs elaborate algorithms to detect high-level features from data facilitating better performance in image processing and classification problems, surpassing traditional machine learning[62,103]. Recently, the PLF field has shown extensive interest in deep learning-based livestock identification and localization[62,104,105]. Future research should further improve the models and networks for both machine learning and deep learning, modify the set of technologies to adapt to these two techniques, and develop automated systems for livestock tracking and health monitoring[100,103].…”