search

bastianwandt / RepNet

This is the original RepNet implementation
86 stars 17 forks source link

Attributes_H36M_2d_3d_training_centralized_17j.mat #4

Open hanhailangya opened 4 years ago

hanhailangya commented 4 years ago

When I get the original data set, how can I convert it to your data form，like:"Attributes_H36M_2d_3d_training_centralized_17j.mat"

zzz0312 commented 2 years ago

Have you solved this problem? If it is solved, can you tell me what is the solution

keliang123 commented 2 years ago

When I get the original data set, how can I convert it to your data form，like:"Attributes_H36M_2d_3d_training_centralized_17j.mat"

Have you converted the data to the .mat format? if you did ,can you tell me how to do it, thank you.

CHIKKIZZY commented 2 years ago

I would like to know this too. The README instructions on training are quite vague and do not address how to organize the data per subject and per activity. We cannot reproduce your results otherwise. Please give a more comprehensive description or share your data preprocessing script. Thanks.

keliang123 commented 2 years ago

I would like to know this too. The README instructions on training are quite vague and do not address how to organize the data per subject and per activity. We cannot reproduce your results otherwise. Please give a more comprehensive description or share your data preprocessing script. Thanks.

hi bro,have you run the train_repnet.py? I hava H36M datasets but there are 32 points. so I have to select the 16 points that is corresponding to the StackedHourglass's 16 points. I found the 16 points [1,2,3,6,7,8,12,13,14,15,17,18,19,25,26,27]from H36M datasets, and I created my training data, but the result is wrong. my problem is whether the points is wrong or the points order is wrong. or my method about creating data is wrong.