Last Saturday I decided to go for a walk and I took my DSLR camera with me. Having watched many Computer Vision and OpenCV videos in my free time lately, I thought it would be a good idea to film the bridge for a couple of minutes and see what I can extract from the video later on during the weekend. So, here is the test setup:
Luckily, a train crossed the bridge during my first take, so I didn't have to wait for another 20 minutes until the next train arrived. To film this video, I used a Nikon D3100 DSLR camera and a low-end 300 mm lens (nothing fancy). I was 1.7 km away from the middle of the bridge:
To analyze the video, I used OpenCV and Python 3. I didn't really think the train would cause any detectable deflection to the bridge, especially since I was very far away, but I was mistaken. Here is what happened to the bridge when the train crossed it:
Here is the result of the measurement in pixels versus time. Currently I don't have enough info to convert pixels into centimeters or inches, but I will contact a few people soon and I hope to find for example the dimensions of the lower chord of the truss. That way, I will be able to convert in an approximate manner pixels into centimeters:
Interestingly, the video was able to detect deflections even before the train entered the bridge. These deflections must be due to a combination of normal traffic and wind.
Several research papers have investigated the possibility of using Computer Vision for Structural Health Monitoring of Engineering Structures. Up to now, I didn't think the technology was so accessible to the structural engineers. I recommend you to take a look at OpenCV and its capabilities.
If you found inspiration in this post, if you liked it or if you think someone you know might benefit from it, I kindly ask you to share this post and comment down below. Stay tuned for more!
Hello, can you please share the source code of this? and if possible your contact details?