Looking to analyze a video of a tennis player and provide data metrics like angle of racquet at the top of the motion and at impact, wrists angle at top versus impact, hip and chest rotation and how they compare, head movement, footwork motions, knee bend metrics etc. Any of this possible?
1 Like
Welcome to the forum PacificBeach99 and a happy good new year 2025 to you. 
I guess you can do this in 3 steps:
1. Create something to process the video frame by frame
Write a program or use a python module/plugin/code piece (if you use python) that either support for example OpenCV, Emgu, or in .net with c# for example ALSO aforge .net on github is a very good source.
2. Use a prompt with ChatGPT
Now you can feed the images to ChatGPT and use this prompt (or your own or modify it, however, whatever you come up with that works best for you):
Prompt:
Analyze a given video of a professional tennis player performing various strokes, such as forehand and backhand groundstrokes, serves, and volleys. Your task is to develop an automated system capable of extracting quantitative data metrics regarding the player's technique. Specifically, focus on analyzing the following aspects:
1. **Racquet Angle**: Measure and report the angle of the racquet at the top of the swing motion and at the moment of ball impact. Provide insights into any variations observed during different types of shots.
4. **Wrist Angles**: Compare the angles of the player's wrists at the top of the stroke versus when making contact with the ball. Discuss potential implications for power generation and shot control.
5. **Hip and Chest Rotation**: Evaluate the degree of hip and chest rotation during the swing phase compared to the follow-through. Assess whether there is a consistent pattern across different shots and what effect this may have on overall power and balance.
6. **Head Movement**: Track the player's head position throughout the swing cycle and note any deviations from optimal alignment. Comment on the importance of maintaining proper head positioning for stability and accuracy.
7. **Footwork Motions**: Document the player's footwork leading up to and during the shot, paying attention to the timing, spacing, and balance. Highlight any notable patterns or areas for improvement.
8. **Knee Bend Metrics**: Quantify the extent of knee bend at the start of the shot preparation and at the point of contact. Discuss how varying degrees of knee bend might affect shot power and precision.
Your analysis should include both qualitative observations and numerical data, supported by visual aids such as graphs, charts, and annotated screenshots from the video footage. Aim to provide comprehensive insights that can help the player refine their technique and enhance their overall performance.
3. Create the result
< Result goes here >.
For instance in table form, etc.
Optional if the result is not good enough:
4. Try this and optimize the process
This should help you to get into a loop to iteration of improving the process and get to a result.
P.S.: It may help to even see if something is really moving from frame to frame using a certain threshold to reduce the amount of video frames to be processed AND start and end at a certain given time.
1 Like
Thank you for your response, makes sense. I like the idea of determining if something is moving from frame to frame to reduce the frames processed. Have you done any testing similar to this? Do you think the analysis would be capable of picking up details like the changes in wrist angle from frame to frame?
1 Like