BumjunPark
The URL for code and pre-trained model for Raw-RGB denoising can be found in the Test section.
Thank you for your interest. If you have any question, please contact me.
import torch from torchvision.models.mobilenetv2 import mobilenet_v2 from torchvision.edgeailite import xnn model = mobilenet_v2(False, False, num_classes=6) dummy_input = torch.rand((1, 3, 224, 224)) model = xnn.quantize.QuantTrainModule(model, dummy_input=dummy_input) checkpoint = torch.load('pretrain.pth') model.module.load_state_dict(checkpoint['model']) This...
Hi, thank you for fast response. I used num_classes=6 when pre-train the model with custom dataset, so I have to use the same option when creating the model. I think...
@mathmanu Thank you for your fast response. It seems like working though the training speed is slow. And I also found that the error doesn't occur while shows the same...
@mathmanu Thank you for your careful suggestion. I'll keep that in mind.