ReverseChoreography
Repository for Reverse Choreograph (Fall 2021-Spring 2022)
data:
Folder containing raw motion data
motion capture:
captureScript_mac, mediaposetest3, python_test, reverse_choreo_capture all perform motion-capture tasks. Note that only the first of these works for macs. Also note that python_test may be deprecated.
Note that switching cameras requires changing the video capture index.
preprocessed:
Folder containing files generated by preprocess.ipynb. For each file in the data folder, 3 files (x, y, and z data) are created here containing position and displacement data for each used body point.
parser_output:
Folder containing a single .csv file where each row corresponds to one data run and the columns contain features extracted from that data run. Generated by parser-update.ipynb
recommendation pipeline:
model.sav is the best regression model obtained by running preprocess.ipynb, parser.ipynb, then regression_model_testing.ipynb
classification_model.sav is the best classification model obtained by performing preprocessing, parsing, and classification training.
recommend.py accepts a csv file containing motion information, allows the user to specify a tempo selected by the dancer, recommends 5 songs using model.sav, and recommends 5 songs using classification_model.sav.
Running:
Note that mac-fullScript.py will run all motion-capture and recommendation tasks in one fluid script for macs.
To successfully run recommend.py or mac-fullScript.py, be sure to log into the Spotify web client and have modified the Spotify client credentials in the file. If told that there is no active device when running a script with the Spotify client open, try first playing a song (can be just for 1 second) and then re-running the script.