Universal Bovine Identification via Depth Data and Deep Metric Learning

Asheesh Sharma,Lucy Randewich,William Andrew,Sion Hannuna,Neill Campbell,Siobhan Mullan,Andrew W. Dowsey,Melvyn Smith,Mark Hansen,Tilo Burghardt

2024-03-30

Abstract:This paper proposes and evaluates, for the first time, a top-down (dorsal view), depth-only deep learning system for accurately identifying individual cattle and provides associated code, datasets, and training weights for immediate reproducibility. An increase in herd size skews the cow-to-human ratio at the farm and makes the manual monitoring of individuals more challenging. Therefore, real-time cattle identification is essential for the farms and a crucial step towards precision livestock farming. Underpinned by our previous work, this paper introduces a deep-metric learning method for cattle identification using depth data from an off-the-shelf 3D camera. The method relies on CNN and MLP backbones that learn well-generalised embedding spaces from the body shape to differentiate individuals -- requiring neither species-specific coat patterns nor close-up muzzle prints for operation. The network embeddings are clustered using a simple algorithm such as $k$-NN for highly accurate identification, thus eliminating the need to retrain the network for enrolling new individuals. We evaluate two backbone architectures, ResNet, as previously used to identify Holstein Friesians using RGB images, and PointNet, which is specialised to operate on 3D point clouds. We also present CowDepth2023, a new dataset containing 21,490 synchronised colour-depth image pairs of 99 cows, to evaluate the backbones. Both ResNet and PointNet architectures, which consume depth maps and point clouds, respectively, led to high accuracy that is on par with the coat pattern-based backbone.

Computer Vision and Pattern Recognition,Artificial Intelligence,Machine Learning

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve

The paper aims to achieve automatic identification of individual cows through depth data and deep metric learning, and provides relevant code, datasets, and training weights to ensure the reproducibility of experimental results. As the scale of farms expands, the imbalance between the number of people and cows makes manual monitoring of individual cows increasingly difficult. Therefore, real-time cow identification is crucial for farm management and precision livestock farming.

The paper proposes a deep metric learning method based on depth data for cow individual identification. This method uses Convolutional Neural Networks (CNN) and Multi-Layer Perceptrons (MLP) as backbone networks to learn a generalized embedding space from body shape features, thereby distinguishing different individuals. This method does not rely on specific breed markings or close-up nose features. Network embeddings are clustered using simple algorithms such as k-Nearest Neighbors (k-NN) to achieve high-precision identification, and new individuals can be added without retraining the network.

Specifically, the paper evaluates two backbone architectures: Residual Neural Network (ResNet) and PointNet for point cloud operations. These two architectures process depth maps and point cloud data respectively, achieving high-precision results on the CowDepth2023 dataset. Among them, the ResNet color backbone network achieved a k-NN identification accuracy of 99.97%, while PointNet's accuracy was 99.36%. The research results indicate that these technologies can identify animals using only top-view depth maps, and the model can rely on similar body surfaces for identification even when there are significant differences in body shape features.