Assessing information and traits of cattle herds with UAVs

In the first article, it was shown that data collected with UAVs in both the Netherlands and Poland could be used with Artificial Intelligence to detect, identify and monitor postures of cattle. The AI models showed a classification accuracy of >95% for detecting cows, ~91% for identifying individual cows and ~88% for obtaining cow postures. These results make camera-mounted drones a promising new technology for monitoring cattle in extensive beef production systems.

In the second article, even more in depth researched on the health status of cattle was performed. Monitoring traits with UAVs such as weight and height of cattle could potentially give farmers insights in the status and health of their herds without visiting them. RGB imagery, videos and LiDAR dataset of UAVs were processed to create 3D models of cattle to estimate weight and height. Over 3 years 25 cattle were monitored. LiDAR showed results for estimating withers height (mean error 6cm) and weight (mean error 38 kg). Estimating weight with 3D models from RGB imagery resulted in a mean error of 62kg. The latter result improved to a mean error of 31 kg when three outliers were removed which exceeded twice the standard deviation. These results suggest that the use of UAVs is promising for estimating body dimensions for extensive cattle herds.

The results of both studies are now published. The first article is published in the International Journal of Remote Sensing of Taylor & Francis under the title ‘Detection, identification and posture recognition of cattle with satellites, aerial photography and UAVs using deep learning techniques’ and the second article is published in the journal Smart Agricultural Technology of Elsevier under the title ‘Estimating body dimensions and weight of cattle on pasture with 3D models from UAV imagery’. Both papers are open-source and are available through the links mentioned below

Abstract article 1:

Mücher, C. A., Los, S., Franke, G. J., Kamphuis, C., 2022. Detection, identification and posture recognition of cattle with satellites, aerial photography and UAVs using deep learning techniques. Published online by International Journal of Remote Sensing: https://doi.org/10.1080/01431161.2022.2051634

To obtain specific information about cattle in extensive production systems, the usual labor intensive work done by the farmer to find and visit cattle herds in large pastures can be replaced by using UAVs. UAVs are capable of assessing traits in cows, like distinguishing individuals and postures. Although these traits and the detection of cattle, do not represent resilience and efficiency directly, these may contain information associated to resilience. We performed a feasibility study of remotely sensed imagery (using datasets from satellites, manned aircrafts, and UAVs), and deep learning techniques to detect, count, identify and characterize posture of individual cows in grassland production systems. With these techniques, we focused on : (1) automatic detection of cattle locations and animal counting; (2) cow postures like standing, grazing or lying; and (3) individual cow identification. Data were collected during three field trials in the Netherlands and Poland. Artificial Intelligence was used to classify the objects (cattle) in the drone imagery. Classification accuracies of >95% were obtained for detecting cows. Accuracies of ~91% were obtained for identifying individual cows, and accuracies of ~88% were obtained for cow postures. These results make camera-mounted drones a promising new technology for monitoring extensive beef production systems.

Abstract article 2:

Los S., Mücher, C.A., Kramer, H., Franke, G. J., Kamphuis, C., 2023. Estimating body dimensions and weight of cattle on pasture with 3D models from UAV imagery. Smart Agriculture Technology 4 (2023) 100167. https://doi.org/10.1016/j.atech.2022.100167

Monitoring traits of cattle with an Unmanned Aerial Vehicle (UAV) could assist farmers in monitoring the status and health of their herd without actually visiting them. This feasibility study presents 3D methods to manually estimate body dimensions, such as height and weight, of individual Holstein Friesian cows with stereo RGB imagery, video and LiDAR data from UAVs. In total, 25 different cows over 3 years were monitored and 4,611 images, ∼10 videos, and a LiDAR dataset were analysed. The methods used were estimating weight by analysing 3D models made with RGB imagery and video from UAV, and estimating height and weight from LiDAR data. Software used in this study to process the UAV data and to extract 3D models are Agisoft Metashape Professional, CloudCompare, RiPROCESS 1.8.4 Riegl Software and Potree Desktop. LiDAR showed accurate results for estimating withers height (mean error 6cm) and weight (mean error 38 kg). Estimating weight with 3D models extracted with Structure for Motion (SfM) from overlapping RGB imagery resulted in less accurate results (mean error 62kg). The latter result improved to a mean error of 31 kg when three outliers were removed which exceeded twice the standard deviation. These results suggest that the use of UAVs is promising for estimating body dimensions in extensive beef production systems. The main challenge remains to record individual cows from all sides without too much movement or motion from the cow to enable 3D modelling. Another challenge was to decrease the mean error in weight estimations so that changes in body weight over time can be monitored. Future research should focus on improving the overall mean error of cattle weight estimation, finding solutions for moving cattle and shifting from manual to automated processing of UAV imagery into 3D models.