1

I am trying to track a tennis ball moving at high speed in a lawn tennis game. I am unable to figure out which approach would be best. I need to perform these detection on a jetson nano(or jetson nano orion).

I have thought of the following approach:

  • Using Stereo camera (placed at a specific point most likely behind one of the player to track incoming ball) for detecting ball using Computer Vision.
  • Using IR base Motion capture system with reflective markers (as used in https://www.youtube.com/watch?v=MHTizZ_XcUM&t=99s)
  • 3D LiDAR
  • Ultrasonic Sensor

The following are the drawbacks of these:

  • Stereo Camera: Requires high shutter speed camera for detecting the ball moving at high speed. This high-end camera is most likely(according to my research) not available for jetson nano or jetson nano orion.

  • IR Based Motion Capture: In the youtube video they were using https://www.vicon.com/hardware/cameras/vero/. Dont know if it’s available in India. Also its cost should be under 20,000 Rupees(240$).

  • 3D LiDAR: Doubtful if it would be able to detect high speed ball.(Best contender till now)

  • Ultrasonic Sensor: Worst accuracy. Low cost is its only selling point.

Let me know if any other options are available.

jcaron
  • 2,408
  • 5
  • 10
Pratham
  • 19
  • 1

1 Answers1

1

Consider using OpenCV and cameras to track tennis balls using computer vision. This github.com project can find tennis balls. For tacking tennis balls this may be a better example to follow.

The computer language Python is often used for quick implementation of applications. And many aspects of Python are written in the computer language C for speed. As Python is 100s if not more times slower than C. If speed it important consider writing your entire OpenCV in a faster language such as C or C++.

st2000
  • 279
  • 1
  • 7