search

zylo117 / Yet-Another-EfficientDet-Pytorch

The pytorch re-implement of the official efficientdet with SOTA performance in real time and pretrained weights.
GNU Lesser General Public License v3.0
5.21k stars 1.27k forks source link

warning #99

Open KelvinHuang666 opened 4 years ago

KelvinHuang666 commented 4 years ago

[Warning] Ignoring Error(s) in loading state_dict for EfficientDetBackbone: size mismatch for classifier.header.pointwise_conv.conv.weight: copying a param with shape torch.Size([810, 112, 1, 1]) from checkpoint, the shape in current model is torch.Size([9, 112, 1, 1]). size mismatch for classifier.header.pointwise_conv.conv.bias: copying a param with shape torch.Size([810]) from checkpoint, the shape in current model is torch.Size([9]). [Warning] Don't panic if you see this, this might be because you load a pretrained weights with different number of classes. The rest of the weights should be loaded already.

zylo117 commented 4 years ago

Don't panic if you see this, this might be because you load a pretrained weights with different number of classes. The rest of the weights should be loaded already.

KelvinHuang666 commented 4 years ago

if i train myown dataset(2500training picture and 500 val picture) with one class .how can i set my params(eg:num_epochs) thanks so much

heitorrapela commented 3 years ago

@KelvinHuang666 You need to search for the best config. of params for your dataset. For the number of epochs I suggest that you put a high number of epochs and keep tracking of the valid loss (add some early stopping to it) or save just the best model during the validation.