I use my own data for retargeting, but the program will give an exception about deep-motion-editing

[PeizhuoLi]

Our architecture requires a consistent topology between training character and test character. Thus, if you want to test with a new topology you will need to retrain the model with your own dataset.

[fryegg]

Hi, I convert example.bvh file to csv file, and I found the number of joint was 31:
Hips Lhipjoint leftupleg leftleg leftfoot lefttoebase rhipjoint rightupleg rightleg rightfoot righttoebase lowerback spine spine1 neck neck1 head leftshoulder leftarm leftforearm lefthand leftfingerbase lefthandindex1 lthumb rightshoulder rightarm rightforearm righthand rightfingerbase righthandindex1 rthumb
As I understand, the number of joints was 23. What difference between those?

[ANYMS-A]

Traceback (most recent call last):
File "eval_single_pair.py", line 104, in
main()
File "eval_single_pair.py", line 92, in main
new_motion = (new_motion - dataset.mean[i][j]) / dataset.var[i][j]
RuntimeError: The size of tensor a (99) must match the size of tensor b (91) at non-singleton dimension 1
Traceback (most recent call last):
File "demo.py", line 46, in
example('Aj', 'BigVegas', '01.bvh', 'intra', './examples/intra_structure')
File "demo.py", line 42, in example
height)
File "/home/wuxiaoliang/docker/newAPP/deep-motion-editing/retargeting/models/IK.py", line 57, in fix_foot_contact
anim, name, ftime = BVH.load(input_file)
File "../utils/BVH.py", line 58, in load
f = open(filename, "r")
FileNotFoundError: [Errno 2] No such file or directory: './examples/intra_structure/result.bvh'

Hello, how can I solve this problem？@ @kfiraberman

create the directories "./examples/intra_structure/" by yourself before you run the program.

[wuxiaolianggit]

I've created the directory your way, but the same error still occurs @crissallan

[wuxiaolianggit]

RuntimeError: The size of tensor a (99) must match the size of tensor b (91) at non-singleton dimension 1，

This is the first exception

[ANYMS-A]

RuntimeError: The size of tensor a (99) must match the size of tensor b (91) at non-singleton dimension 1，

This is the first exception

Ah~ Since the tensor's shape doesn't match, I think you should check whether "01.bvh" is the mocap data of “Aj” or "BigVegas".
Because the shape of the motion tensor depends on the number of joints of the character. e.g. 91 = (23-1) * 4 + 3 because "Aj" and "BigVegas" both has 23 joints as its simplified skeleton.

So if tensor a's shape is 99, I would reckon that this bvh file might belongs to a character which has 25 joints， since 99 = (25-1)*4 +3

[wuxiaolianggit]

OK, thank you very much for your reply,

"Our approach is also quantitatively evaluated on a synthetic dataset that contains pairs of motions applied to different skeletons. To the best of our knowledge, our method is the first to perform retargeting between skeletons with differently sampled kinematic chains, without any paired examples. " , it is from "Skeleton-Aware Networks for Deep Motion Retargeting",

Is it required that the number of key points must be consistent？

[wuxiaolianggit]

thank you very much for your reply @PeizhuoLi