search

simochen / flowtrack.pytorch

Pytorch implementation of FlowTrack (Simple Baselines for Human Pose Estimation and Tracking).
74 stars 12 forks source link

dimension match problem #6

Closed 7color94 closed 5 years ago

7color94 commented 6 years ago

Thanks for your code. However, I meet the following problem, could you help me ?

Traceback (most recent call last): File "preprocess.py", line 40, in <module> flow_im = estimate(model, previous, current) File "/data00/home/xxxx/flowtrack.pytorch/tools/flownet/demo.py", line 102, in estimate pred_flow = model(ims_v).cpu().data File "/data00/home/xxxx/.pyenv/versions/anaconda3-4.0.0/lib/python3.5/site-packages/torch/nn/modules/module.py", line 491, in __call__ result = self.forward(*input, **kwargs) File "/data00/home/xxxx/flowtrack.pytorch/tools/flownet/../../lib/flownet/model/models.py", line 271, in forward concat5 = torch.cat((out_conv5,out_deconv5,flow6_up),1) RuntimeError: invalid argument 0: Sizes of tensors must match except in dimension 1. Got 24 and 23 in dimension 2 at /pytorch/aten/src/THC/generic/THCTensorMath.cu:111

simochen commented 6 years ago

Hi, @7color94 . What's your input image size?

7color94 commented 6 years ago

Two input images are all (720, 1280, 3).

simochen commented 6 years ago

@7color94 This problem is caused by the stride2 convolution and deconvolution after 'out_conv5' layer. You can padding input image and make input size divisible by 64.