Dairy Cow Tiny Face Recognition Based on Convolutional Neural Networks

Abstract

In practical applications of cow face recognition, the accuracy is often lower than expected because of the influence of camera's low resolution and position. In this paper, we aim to develop and pilot a method for improving recognition accuracy and recovering identity information for generating cow faces closed to the real identity. Specifically, our network architecture consists of two parts: a super-resolution network for recovering a high-resolution cow face from a low-resolution one, and a face recognition network. The superresolution network is cascaded with the recognition network. An alternately training strategy was introduced to ensure the stability of the training process. The cow face dataset was collected by us, which contains 85200 dairy cow face images from 1000 subjects. Experimental evaluations demonstrate the superiority of the proposed method. Our method has achieved 94.92% recognition accuracy on the small size (12 Â 14) cow face.

A Systematic Review of Machine Learning Techniques for Cattle Identification: Datasets, Methods and Future Directions

“…The body coat pattern and face are unique features for identifying individual cattle (Kumar and Singh, 2020). With the recent advent of ML and DL methods, facial and body coat patterns have been widely used for cattle identification (Arslan et al, 2014;Andrew et al, 2016;Yang et al, 2019). For the ML models, the features are extracted from the face or body images and then fed into the models for identification.…”

“…As it requires a long time to process the high-quality image datasets, the researchers reduced the image size or divided the images into small pieces to increase the data processing speed (Yao et al, 2019;Yang et al, 2019). Several studies in this SLR report that small and unbalanced datasets are the causes of low accuracy (Kumar et al, 2017a(Kumar et al, , 2018aQiao et al, 2019;Guan et al, 2020).…”

Cow Face Recognition for a Small Sample Based on Siamese DB Capsule Network

“…In (5), is the distance between two features, and Y is the label of the image pair. When Y = 1, the two images belong to the same category, and L minimizes the distance between the two features.…”

“…For cattle facial identification, the super-resolution network was applied as a precursor network. While conducting image recognition, the image information was recovered [5]. Bisen used the K-means clustering algorithm to constantly deal with six prior frames and then used YOLO3 to detect cow faces.…”

“…It causes the cow faces in the photographs to have various orientations. In addition, the light, background, and other variables vary substantially [3]- [5]. Therefore, cow face data gathering is difficult, and identification is costly.…”