RuntimeError while Training on Custom Dataset

[Bowenwu1]

Hi! Thanks for your wonderful work!
For now, I'm trying to train impersonator on my data. But I got the error as ablow and have no idea to solve it. I think it may be some issue related to my data(may SMPL). Could you give me some advice to figure it out?
Looking forward to your reply! You will help a lot for my research.

File "train.py", line 140, in <module>                                                                                 [16/1919]
    Train()
  File "train.py", line 26, in __init__
    self._train()
  File "train.py", line 39, in _train
    self._train_epoch(i_epoch)
  File "train.py", line 68, in _train_epoch
    self._model.optimize_parameters(keep_data_for_visuals=do_visuals, trainable=trainable)
  File "/data0/wubowen/impersonator/models/impersonator_trainer.py", line 354, in optimize_parameters
    loss_G = self._optimize_G(fake_bg, fake_src_imgs, fake_tsf_imgs, fake_masks)
  File "/data0/wubowen/impersonator/models/impersonator_trainer.py", line 385, in _optimize_G
    fake_tsf_imgs, self._real_tsf, bbox1=self._head_bbox, bbox2=self._head_bbox)) * self._opt.lambda_face
  File "/data0/wubowen/miniconda3/envs/liquid/lib/python3.6/site-packages/torch/nn/modules/module.py", line 547, in __call__
    result = self.forward(*input, **kwargs)
  File "/data0/wubowen/miniconda3/envs/liquid/lib/python3.6/site-packages/torch/nn/parallel/data_parallel.py", line 152, in forwa$
d
    outputs = self.parallel_apply(replicas, inputs, kwargs)
  File "/data0/wubowen/miniconda3/envs/liquid/lib/python3.6/site-packages/torch/nn/parallel/data_parallel.py", line 162, in paral$el_apply
    return parallel_apply(replicas, inputs, kwargs, self.device_ids[:len(replicas)])
  File "/data0/wubowen/miniconda3/envs/liquid/lib/python3.6/site-packages/torch/nn/parallel/parallel_apply.py", line 85, in paral$
el_apply
    output.reraise()
  File "/data0/wubowen/miniconda3/envs/liquid/lib/python3.6/site-packages/torch/_utils.py", line 369, in reraise
    raise self.exc_type(msg)
RuntimeError: Caught RuntimeError in replica 3 on device 3.
Original Traceback (most recent call last):
  File "/data0/wubowen/miniconda3/envs/liquid/lib/python3.6/site-packages/torch/nn/parallel/parallel_apply.py", line 60, in _work$
r
    output = module(*input, **kwargs)
  File "/data0/wubowen/miniconda3/envs/liquid/lib/python3.6/site-packages/torch/nn/modules/module.py", line 547, in __call__
    result = self.forward(*input, **kwargs)
  File "/data0/wubowen/impersonator/networks/networks.py", line 246, in forward
    head_imgs1 = self.crop_head_bbox(imgs1, bbox1)
  File "/data0/wubowen/impersonator/networks/networks.py", line 306, in crop_head_bbox
    head = F.interpolate(head, size=(self.height, self.width), mode='bilinear', align_corners=True)
  File "/data0/wubowen/miniconda3/envs/liquid/lib/python3.6/site-packages/torch/nn/functional.py", line 2503, in interpolate
    return torch._C._nn.upsample_bilinear2d(input, _output_size(2), align_corners)
RuntimeError: input and output sizes should be greater than 0, but got input (H: 0, W: 45) output (H: 112, W: 96)

[Bowenwu1]

I extract SMPL data as you have shown in #28.