Hey, I’m Jacob. I am newer to this forum as a contributor; previously, I was more of a reader.
I am a dairy farmer and I started with other farmers an open-source project called CowCatcher AI. In short: CowCatcher AI is an open-source computer vision system designed to monitor your herd 24/7. By analyzing live footage from your barn cameras, it automatically detects “mounting” behavior—the primary sign of estrus (heat)—and instantly sends a photo notification to your smartphone via Telegram, or you can combine it with your Home Assistant setup.
With a simple setup using a standard computer and a “cheap” IP camera in your barn (cubicle or free stall, it doesn’t matter), you are notified if a cow mounts another cow. We send the picture and a short video to you, so you can recognize the cow and decide if you are going to inseminate or, for example, just log that a steer fertilized a cow. There is a very broad way of how it can be used.
The project started in March 2025 as an idea. What started with the idea to detecting only for young heifers for their first insemination 13 to 15 months, became fast for the whole barn. Within 3 months, we already had a simple, cheap, and effective solution, the first calf’s detected with cowcatcher are now getting born, and farmers are getting good results, I find it performing more accurate with the time than the activity collars I already had; I now solely rely on CowCatcher and just use those as a backup. The advantage is you can see the heat in progress, first you see the cow in heat mount other cows after couple hours, he gets mounted and becuase of the first mount you can count 16 to 24 hours to inseminate on the perfect time. What activity cant pinpoint as accurate.
Currently, we are effectively working with 2 or 3 other dairy farmers on the codebase. I hope to have an “easy to install” web app with all the bells and whistles ready by March. We already have a production-ready program that is running on 20+ farms in Belgium, the Netherlands, and France to my knowledge. It has been downloaded over 100 times, so it’s already bigger than I thought.
We hope to grow the community with users. Through close communication between users and us developers, we can turn ideas and feedback into something truly beautiful. It’s quite a process with mistakes and solutions that we thought would work but didn’t, but it’s getting better and better.
Question feel free to ask or want to start with Cowcatcher AI yourself and need help mail to [email protected]
Github: https://github.com/CowCatcherAI/CowCatcherAI Video (in Dutch, but you get the idea): https://www.youtube.com/watch?v=f7Cmz12qZJ0