question

[suqian1988]

Can the tennis-tracking mark the ball bounce position in the video?

thank you very much.
Mike

[shukkkur]

@suqian1988 Hi Mike, thanks for opening the issue!
Yes, it sure can.
Here, if you pass --bounce=1, it should mark the bounces in blue and display them, like in the following image (unless the model failed to predict bounces 😅)

If you want all the bounces to be marked on the court, then 👇

tennis-tracking/predict_video.py

Lines 337 to 339 in 91f0fa2

	predcted = clf.predict(X)
	idx = list(np.where(predcted == 1)[0])
	idx = np.array(idx) - 10

the list idx stores the frame numbers where the ball touches the court. So, change this

tennis-tracking/predict_video.py

Lines 357 to 370 in 91f0fa2

	while True:
	ret, frame = video.read()
	if ret:
	# if coords[i] is not None:
	if i in idx:
	center_coordinates = int(xy[i][0]), int(xy[i][1])
	radius = 3
	color = (255, 0, 0)
	thickness = -1
	cv2.circle(frame, center_coordinates, 10, color, thickness)
	i += 1
	output_video.write(frame)
	else:
	break

with this

  while True:
    ret, frame = video.read()
    if ret:
      # if coords[i] is not None:
      for loc in idx:
        center_coordinates = int(xy[loc][0]), int(xy[loc][1])
        radius = 3
        color = (255, 0, 0)
        thickness = -1
        cv2.circle(frame, center_coordinates, 10, color, thickness)
      i += 1
      output_video.write(frame)
    else:
      break

Didn't run the code myself, but should work.
Feel free to reopen the issue, anytime!
😄

[suqian1988]

oh,on the small black court, right? is there a way to mark the bounce points on the main court? thank you so much. Mike.

…

On Mon, Feb 14, 2022 at 7:11 AM Shukur Sabzaliev ***@***.***> wrote: @suqian1988 <https://github.com/suqian1988> Hi there, thanks for opening the issue! Yes, it sure can. In README.md I included a gif image where the bounces are *marked in blue*, here [image: image] <https://user-images.githubusercontent.com/78250180/153860050-72db9991-b505-4e7d-bebb-253f926a419c.png> If you want all the bounces to be marked on the court, then 👇 https://github.com/ArtLabss/tennis-tracking/blob/91f0fa2ba740e9840d23c5c0bf3dbeb3b62cbe05/predict_video.py#L337-L339 the list idx stores the frame numbers where the ball touches the court. So, change this https://github.com/ArtLabss/tennis-tracking/blob/91f0fa2ba740e9840d23c5c0bf3dbeb3b62cbe05/predict_video.py#L357-L370 with this while True: ret, frame = video.read() if ret: # if coords[i] is not None: for loc in idx: center_coordinates = int(xy[loc][0]), int(xy[loc][1]) radius = 3 color = (255, 0, 0) thickness = -1 cv2.circle(frame, center_coordinates, 10, color, thickness) i += 1 output_video.write(frame) else: break Didn't run the code myself, but should work. Feel free to reopen the issue, anytime! 😄 — Reply to this email directly, view it on GitHub <#13 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGEJ6UWQI653EAP6G54SSILU3DWNHANCNFSM5OHCONCQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[shukkkur]

@suqian1988 No, the code provided will mark the bounces only on the real court itself.
As for for marking on the black court (mini-map), it is a bit more complex. We will need the inverse matrix and will then apply perspective transform.