mks0601 / I2L-MeshNet_RELEASE

Official PyTorch implementation of "I2L-MeshNet: Image-to-Lixel Prediction Network for Accurate 3D Human Pose and Mesh Estimation from a Single RGB Image", ECCV 2020
MIT License
724 stars 127 forks source link

RuntimeError: Error(s) in loading state_dict for DataParallel: #133

Open baboonaaaa opened 2 months ago

baboonaaaa commented 2 months ago

I find the following troubles when I compile demo.py,can you tell me how to resolve it?

size mismatch for module.pose_net.conv_z_1.0.weight: copying a param with shape torch.Size([16384, 2048, 1]) from checkpoint, the shape in current model is torch.Size([101376, 2048, 1]). size mismatch for module.pose_net.conv_z_1.0.bias: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]). size mismatch for module.pose_net.conv_z_1.1.weight: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]). size mismatch for module.pose_net.conv_z_1.1.bias: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]). size mismatch for module.pose_net.conv_z_1.1.running_mean: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]). size mismatch for module.pose_net.conv_z_1.1.running_var: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]). size mismatch for module.pose2feat.conv.0.weight: copying a param with shape torch.Size([64, 1920, 3, 3]) from checkpoint, the shape in current model is torch.Size([64, 11548, 3, 3]). size mismatch for module.mesh_net.conv_z_1.0.weight: copying a param with shape torch.Size([16384, 2048, 1]) from checkpoint, the shape in current model is torch.Size([101376, 2048, 1]). size mismatch for module.mesh_net.conv_z_1.0.bias: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]). size mismatch for module.mesh_net.conv_z_1.1.weight: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]). size mismatch for module.mesh_net.conv_z_1.1.bias: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]). size mismatch for module.mesh_net.conv_z_1.1.running_mean: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]). size mismatch for module.mesh_net.conv_z_1.1.running_var: copying a param with shape torch.Size([16384]) from checkpoint, the shape in current model is torch.Size([101376]).

mks0601 commented 2 months ago

what did you change from the original repo?

baboonaaaa commented 2 months ago

Snipaste_2024-09-08_21-22-51 thanks for replying,but I'm not change it