Hello i want to calculate some metrics and decissions when i get a good recognition of player and ball (in first instance). After that i want to know the best form to create a Dataset. The idea i use a video and recognize and player in real time of the video. Is it necessary to retrain the model if the scenario of the video-game changes? is available datasets online for this purpose?
Hi @Alex_Alzueta!
First off, welcome to the Roboflow community!! To help me best understand your question, can you help me understand how the scenario of the video game may change? An example would be incredible helpful.
There are over a million datasets available on universe.roboflow.com, and once I fully understand your purpose, I’m happy to point you at some specific datasets!!
yes, different courts could be present….basically local leagues, teams etc..….at first time i want to calculate metrics and the root of these metrics are for example which team has got the ball(possesion) in one instant of the game (clip) . I dont need to be in Real time only on video should be processed. A full curt or mid court would be neccesary
This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.