Technical Note: Estimating Cow Live Bodyweight Using Alternative Sensing

Yijie Xiong Yijie Xiong, Precision Livestock Management Extension Specialist

Obtaining reasonable cow body weight is an important indicator of production efficiency and offers the potential to inform producers if the animals experience any health issues. Although “weighing” the animals sounds old-school and an easy task, it is not as convenient as it sounds. In fact, in some circumstances, it is rather difficult to acquire accurate live bodyweight for these large animals, especially if you do not have the proper tool (i.e., hydraulic chute) to perform the task. Starting Spring 2021, we began a research project at the Gudmundsen Sandhills Lab to estimate cow live bodyweight using an alternative sensing tool – depth camera. 

Why is it important to know your cows’ weight?

Depending on the breed and age, an average mature beef cow can weigh anywhere between 900 lb. to over 1,500 lb. Maintaining proper body weight not only provides insights regarding their reproductive health, but rapid changes in body weight may also indicate early signals for disease. Currently, the industry’s gold standard is to use a hydraulic squeeze chute combined with a load cell to obtain cattle body liveweight by temporarily restraining an animal in the chute. The hydraulic squeeze chute can cost anywhere between $11,000 to $20,000, or even more depending on the model and manufacturer. This price does not include the necessities such as the walking alleys and the load cell. 

What is a depth camera and how does it work?

A depth camera can sometimes be referred to as a “3-D camera”. By definition, it captures the objective in three dimensions. In addition to the 2D plane, the depth camera also measures the distance from the object of interest to the lens. It may sound like a tech guru term, but if your children or yourself play sports games via popular game consoles such as XBOX, Wii, or PlayStation, you are no strangers to the depth camera – those consoles utilize exactly the same technology as we are deploying on cows here! There is already work being done using different types of depth sensors to estimate body weight, body condition scores, or even detect gait patterns for animals, but limited to other species (pigs and dairy cattle) or other locations (Europe and Australia). Unfortunately, not much work has been established for US beef cattle. Therefore, we wanted to evaluate the feasibility of using depth cameras to estimate beef cattle live bodyweight. 

Prior to data collection, we placed the depth camera above the walking alley and beneath the roof and positioned it facing down the walking alley. During the data collection, we had cows enter the alley in a smaller batch (2 to 3) and take top-view depth images for each animal. Since this is to validate the method, we also captured the load cell measured cattle weights. Post data collection, we used a customized computer analytic program (MATLAB) to process the images, eliminate background pixels, and extract pixels that represent the approximate volume of the cows from the images. We then develop a linear regression model between the image-estimated body volume and the scale-measured weights.

How does this alternative tool relate to producers?

We are still a few steps away from developing the prototype for a fruitful application of this alternative sensing tool. However, as discussed above, obtaining accurate cow body weight can be unpractical and cost-prohibitive for both small-size producers and extensive ranches. If successful, this tool can serve as a cost-saving but precise tool for obtaining cow body weights with reasonable accuracy. I vision that such tools will be deployed at the waterers to take images of cows/calves and provide weight estimates in the future.

Depth Image

Figure 1. An example of color adjusted depth image taken in front of a beef cow at GSL.