Training issues with gaitgraph1

[fffaaaaaa]

class JointNoise(object): """ Add Gaussian noise to joint std: standard deviation """

def init(self, std=0.25): self.std = std

def call(self, data):

T, V, C

noise = np.hstack((
    np.random.normal(0, self.std, (data.shape[1], 2)),
    np.zeros((data.shape[1], 1))
)).astype(np.float32)

return data + np.repeat(noise[np.newaxis, ...], data.shape[0], axis=0)
报错
Traceback (most recent call last):

File "D:\gait recognition\OpenGait-master_new\opengait\main.py", line 72, in run_model(cfgs, training) File "D:\gait recognition\OpenGait-master_new\opengait\main.py", line 54, in run_model Model.run_train(model) File "D:\gait recognition\OpenGait-master_new\opengait\modeling\base_model.py", line 406, in run_train ipts = model.inputs_pretreament(inputs) File "D:\gait recognition\OpenGait-master_new\opengait\modeling\base_model.py", line 310, in inputs_pretreament seqs = [np2var(np.asarray([trf(fra) for fra in seq]), requires_grad=requires_grad).float() File "D:\gait recognition\OpenGait-master_new\opengait\modeling\base_model.py", line 310, in seqs = [np2var(np.asarray([trf(fra) for fra in seq]), requires_grad=requires_grad).float() File "D:\gait recognition\OpenGait-master_new\opengait\modeling\base_model.py", line 310, in seqs = [np2var(np.asarray([trf(fra) for fra in seq]), requires_grad=requires_grad).float() File "D:\Users\15840\anaconda3\envs\py39\lib\site-packages\torchvision\transforms\transforms.py", line 94, in call img = t(img) File "D:\gait recognition\OpenGait-master_new\opengait\data\transform.py", line 339, in call return data + np.repeat(noise[np.newaxis, ...], data.shape[0], axis=0) ValueError: operands could not be broadcast together with shapes (60,64,44) (60,64,3) 有大佬知道怎么修改吗

[jdyjjj]

GaitGraph1 is a pose based method. It seems that you use silhouette dataset. You can see how to pretreat the pose dataset. Good luck.

[fffaaaaaa]

Thank you