From CCTV to AI — How Computer Vision Detects Stress Before Humans Can

I’m often asked: “Do we really need AI when we already have cameras?”

My answer is always the same: cameras that only record are wasted assets.

Most farms already capture video. The problem is that no one can realistically watch 24 hours of footage per camel, per day. During this project, we transformed passive CCTV into an active monitoring system—one that doesn’t get tired, distracted, or biased.

What struck me most was how early stress behaviors appeared. Distressed sitting, repeated rope pulling, subtle agitation—these showed up hours or days before anything a human would flag as a problem.

I’ve seen similar patterns in industrial safety systems: accidents are preceded by small signals. Animal welfare is no different. Early detection is not a luxury; it is the only window where prevention is still possible.

AI didn’t make decisions for us. It simply showed us what we were missing.