The developers of a visual tracking system that identifies lameness in dairy cows is developing the technology for body scoring as well.
OmniEye uses a camera on the exit race of a dairy shed to monitor cow movement by filming and analysing her stride every time she leaves a cowshed.
Artificial intelligence (AI) then detects and scores any lameness – and the technology could soon be used provide condition scoring of cows.
David Newstead, OmniEye operations manager, was at the recent South Island Dairy Event in Invercargill and said lameness scores are based on DairyNZ’s measurement.
To operate efficiently, a 6m-long, flat exit race is required, leading to a post on which a camera and electronic identification (EID) reader are mounted.
The camera collects images, picking up any changes to her gait.
If there is an issue, a video, lameness score and tag number are sent to managers or staff.
The cause could be as insignificant as a stone in her hoof or a more serious onset of lameness.
Vets have been involved in training AI to identify and interpret the data and images.
Operating on 40 farms, the OmniEye technology is becoming more accurate, Newstead said, able to identify cows by body markings or facial recognition.
Newstead said it is up to farmers to use the information on potential lameness.
“We’re the early detection piece. There is no point having us if you’re not going to treat her.”
The returns are worthwhile.
A lame cow is estimated to cost over $500 in direct and indirect costs.
Over half of that is through lost milk production with the balance vet and treatment costs, culling, discarded milk and the impact on reproduction.
It is a bigger problem in South Island herds, where an estimated 26% of cows become lame each year, compared to the North Island, where the figure is 16%.
Newstead said the next step is to incorporate into the technology hoof pathology using AI, such as information on which hoof is lame, if it is a repeat issue and any previous treatment she has undergone.
This could provide a warning of trends of recurring problems.
For example, if a farmer has a tolerance lameness threshold score of 1.5, a cow moving from 1 to 1.5 would not be considered an issue.
OmniEye technology could be trained to identify cows with progressively worsening lameness on successive days, providing an alert before she reaches the threshold.
Similarly, that information can be cross-correlated to other events such as wet weather or paddocks grazed, potentially providing a link with lameness.
Now OmniEye wants to use that technology to assist with condition scoring.
Newstead said body scoring cows is easier because it requires only static 3D images using technology that already exists.
“We’ve got the technology and the computer infrastructure.”
