alcantarar / recurrent_grf_prediction Goto Github PK

I have a question regarding setting up the data needed for GRF prediction based on user specific data. Regarding the acceleration data such as Sample_test_data.csv, I would like to know how that should be formatted. Is there a specific part of the leg that should be identified as the shank and what are the formatting requirements for the .csv. Regarding formatting requirements is there a sampling rate range, can multiple shanks be added...etc? Regarding footstrike I was hopping you could let me know how to create a footstrike file that the program can use based of acceleration data, a video, and subject information. Lastly, how can I get the grf data to be outputted as a excel file, or .trc, or csv?

In summary I basically would like to know how to format my files and data for the program if I have raw acceleration data, a video of the movement for two angles, and subject info like height and weight. I'd also like to know how to get the grf prediction results in a raw data format like excel, csv, or .trc.

Thank you for your time.

Users interested in using the pre-trained model for their own purposes may run into the following roadblocks:

• Making sure data is in wide format for keras
• Windowing their input signal as described in the manuscript
• Extracting features correctly
• Organizing input features (3d array)
• Normalizing signals to min/max based on manuscript data
• Filtering and general paper method replication

Need to add code used to generate figures in manuscript.

Notably Accelerometer.csv instead of Accelerometer_one_sub.csv. Need to check repo contents upon publication.