GoPro videos are an exciting way to capture motion events. Since I can’t help thinking about physics, I wondered if it would be possible to analyze the motion to determine speed and acceleration. I wanted to figure it out so that every time I did something fun and recorded it, I would be able to go back and see how fast I was going or how great my acceleration was.

I found that using these videos to analyze motion using LoggerPro creates some problems.

Traditionally, the videos we have used for analysis are recorded with a stationary camera and a moving object. The fixed background provides a reference frame through which an object moves. But with the GoPro, the camera seems stationary while the world flies past.

I took a GoPro video while riding my bike on a sunny day. In this way, I had a second measure for the bike speed. In the shadow, you can see the wheel going around. So I decided to compare the speed of the bike using data from the shadow and data from the video to see if using video analysis was reliable.

**Speed Calculation – Method 1 – Circumference and Time**

The GoPro records 30 frames per second. If you watch the tire stem as it travels around, you discover that it takes 22 frames or 0.733 seconds to complete one rotation. During this time the bike travels forward one circumference.

The diameter of the bike wheel is 65 cm and circumference of 2.042 m.

This yields a speed of 2.79 m/s using the circumference and time for one rotation.

**Speed Calculation – Method 2 – Video Analysis**

As with all video analysis, some object of known size is used to determine the scale. A meter-stick serves as a good measure as long as it is placed the same distance from the camera as the moving object. If it is not, errors arise. (See previous post). The camera and the reference meter-stick remain the same distance apart through out the video.

The GoPro video is different. All objects in the field move relative to the camera. To overcome this aspect of the video, I chose to do the analysis within a small area directly in front of the camera. The GoPro has a fixed focus of 4 ft. In the bike video analyzed here, the area directly in front of the bike wheel was chosen. At the start of the video I found a feature of known length (85 cm) and was perpendicular to path of motion.

I chose a spot in the road that was moving directly toward me (that I was moving directly toward?) and recorded its position in 8 frames. The data looks fairly straight on the graph, so the slope of the line should represent speed.

This graph shows the slope of the line if all 8 points are included. The slope, and therefore the speed is 2.451 m/s. That result is a little disappointing. Recall that the speed from method 1 was 2.785 m/s.

After some consideration, I decided that part of the problem was that the farther away from the camera the spot in the road was, the less correct the reference distance was. If that were true, the more points included, the less correct the results would be.

If I only included the last 4 points in the analysis, the speed was determined to be 2.698 m/s. That was somewhat better!

If only the last 3 points are used the speed is determined to be 2.791 m/s.

So, the agreement between the two methods improves as fewer points are used, perhaps because the reference distance is more true for these later points (when the spot is the road was nearer to the bike).

I would not usually be in favor of ignoring some data points, but in this case there is a clear reason for doing so. This would be a good conversation to have with students, when would such a decision be legitimate?

I hope this helps you think about some of the issues associated with using video analysis with GoPro videos. This situation was very contrived. The road at least was always a fixed distance from the camera. I imagine that if you jumped off a cliff, nothing would remain constant and the analysis would be even more complex. It does, however, provide another opportunity to talk with students about frame of reference.

Always, your helpful comments are welcome.