Okay, so I needed a way to count how many people were coming through our booth at these trade shows. We used to just kinda guess, or have someone stand there with a clicker, but that was super unreliable and, honestly, a waste of a person’s time. So, I started looking into a better solution.

My DIY Foot Traffic Counter Journey

First, I brainstormed a few ideas. I thought about maybe using some kind of pressure mat, but those seemed expensive and like they might break easily. Then I considered those laser tripwire things, but I worried about people tripping or messing with them. Finally, I landed on the idea of using a camera and some computer vision stuff.

I grabbed an old webcam I had lying around – nothing fancy, just a basic one. Then I started digging around for software that could do what I needed. I am not very good with code, so I was looking for premade solution.

I spent a good few hours messing around with different settings and camera positions. I learned that lighting is super important. If it’s too dark, the software can’t “see” people properly. If it’s too bright, you get weird shadows and reflections that mess things up. I ended up adding a small lamp near the booth entrance to get the lighting just right.

I tested the whole setup by walking back and forth in front of the camera a bunch of times. I compared the software’s count to my manual count, just to make sure it was accurate. It wasn’t perfect at first – sometimes it would count me twice if I walked too slowly, or miss me completely if I went too fast. But after some tweaking, I got it working pretty reliably.

Finally, I took the whole setup to a small local exhibition to see how it would work in the real world. I set up the camera above our booth entrance, made sure the lighting was good, and let it run all day.

  • It was so cool to see the numbers going up throughout the day!
  • I could even see peaks and dips in the traffic flow, which was super interesting.

At the end of the day, I checked the final count, and It gives us real, hard data to work with, instead of just guessing. I am very happy about this.