Closed JeanneBod closed 1 year ago
Hey @JeanneBod !
I created this .csv
manually using 9 tennis videos (rallies) and marked the coordinates of bounce (ball touching the court).
Then, before feeding this data to a time series model (sktime
models) I feature engineered it a bit. Please refer to the following notebook to see all the steps applied to the dataset (SKTIME section of the noteboo)
@jaehyunshinML Hey there! Here is the complete notebook, I guess you can find the training code for the bounce in
SKTIME
section of the notebook. Good Luck! And for the record, it's mainly about transforming the dateset into the right format forsktime
Feel free to reopen the issue if you will have any additional question :) Best of Luck!
In the bigDF.csv, how did you define the bounce? Is it an observed bounce by an observator visualising of the video? Thank you for your answer