What should i do if i want to evaluate DMNN on my own skeleton database?

[FesianXu]

Hi, what should i do first if i want to evaluate this algorithm on my new collected skeleton database? Do i need to modify a lot of code? Any suggestion will be appreciated :)

[FesianXu]

Also, i notice that you have used the ConvLSTM to manage the spatial-temporal infomation which seems to be not reported in the paper, i am curious about it. Would you mind to report the perfomance of it compared to the 3D Conv one? Thanks