Cowmon!

We conducted research on the dangers of transportation. Mortality rates were recorded from 0.1% to 0.01% between 2010 and 2015. This means thousands of animals die every year just in the European Union. This is a significant number both from an animal welfare perspective and a financial perspective. In addition to this, injuries are even more common, which also cause a lot of suffering to the animals as well as extra costs to the owners. As a solution, we came up with the idea to monitor the cows during transport. We install cameras whose images are analyzed by artificial intelligence. If it detects an abnormality, such as a cow being trampled, it alerts the driver, giving him the opportunity to intervene.

From the posture of the cows, it can be deduced with great certainty whether they are in danger.

Our solution is primarily made to help avoid trampling incidents, but it can also help reduce the impact of other problems (e.g.: shipping fever).

In the event of an alarm, the driver has the option to stop and intervene or even seek assistance from an expert. Quick intervention can save a significant portion of animals in danger.

By connecting the solution to an existing telemetry system, a central operator team with the right expertise could control the transportation. Using the alarms generated by the artificial intelligence, or via live camera view or replays, the driver can be given appropriate instructions. This can make the intervention more effective and reduce the impact of AI false positive alerts.

We have looked at how animals are currently transported and what rules need to be followed.

There are transport companies that are already using camera surveillance. But while driving, the driver can’t pay attention to both the road and the camera image at the same time, so this is only used for the occasional ‘glance’. Our idea is to have the artificial intelligence analyze the image, making transportation easier and more secure.

We wanted to create a demo solution, with which we can illustrate the process and gather experience about the way artificial intelligence operates and its challenges. It was easier than we thought. From easily available solutions and tutorials on the Internet, we’ve been able to put together a solution that recognizes cows from a top view and alerts the driver when it detects a cow with an abnormal posture (lying down).

Our application, COWMON! got its name by putting together the words cow and monitoring. We also designed a logo for the product.

Our goal was for the application to detect the location of the animals, mark the (abnormally) lying animal in the image, and send an alert if it finds one.

After selecting the framework, we took pictures of our model cows from multiple angles and positions. We trained the AI with the finished photos for about 10 hours.

Even now, we are not sure that basing the model on object recognition is the correct solution. We thought about using a combination of motion-detection and a wireframe-based solution, but we decided to use object recognition for the demo because of its simplicity.

We’ve tried to check what technologies are on the market for similar purposes, and so far we’ve only come to the conclusion that “It can work”.

During the creation and testing of the model, we found that it is much easier to work with artificial intelligence than we previously thought and that a lot of data (images) would be needed to develop a solution that works accurately on a larger scale.

Impact

Our idea is a solution to a big animal welfare problem.

There are already transport companies who use camera surveillance, but we plan to analyze the image with artificial intelligence, so it won’t distract the driver. During long-term transport, even on a ship, it stays “watchful” all the way through. It also eliminates problems arising from negligence, inattention, or lack of expertise.

During our research, we received a lot of feedback from the interviewed professionals that our solution could help some of their problems. They mentioned specific cases where an accident could have been avoided with our solution.

Although the direct economic benefit can be demonstrated in the long run, in our experience (also beyond the scope of this research project) this does not provide sufficient motivation for the companies to develop it, and replace the already working, “good enough” methods.

The regulations of the European Union set quite strict standards during transport. These are expected to be further tightened. The best way to spread the solution would be for regulations to evolve in that direction. We plan to contact the relevant authorities in this regard.

Several shipping companies we contacted during our research would be interested in the solution. If the product were available in a complete, “boxed” form, these companies could easily integrate it.

COWMON! would be useful both from an economical and an animal welfare point of view, breeders, slaughterhouses, and transporters would benefit from using it, but perhaps the biggest benefiters would be cows

One of the biggest challenges would be the placement of the cameras.

There is not much space around / above the animals, so a wide-angle solution would have to be used. A glass enclosure would be adequate for the physical protection of the devices due to its durability and cleanability.

We have consulted with a shipping company about their experience in using the camera system and based on this we deemed the task to be doable.

It is difficult to estimate, but based on our information so far, the development of a professional finished product would require 800-1000 engineering days (4 FTE, 1 year) of development work. We estimate this to be between € 150,000 and € 200,000, which means a very low license cost even for only 1,000 vehicles.

Cost elements of equipping a transport vehicle:

• 5 cameras (wide angle, with glass cover)

• on-board module (storage, analysis)

• installation, cabling

Based on data available on the Internet, we estimate this to be 1500-2000 EUR / vehicle, which is a small investment compared to the price of the vehicle and the cost of other equipment.