There are doubts about the CNN features of the image dataset mentioned in your article. Are you using a base model (such as torchvision.models.resnet18(pretain=True) or others) to extract features or a custom CNN module? In my experiment, I found that different feature extractions still have a great impact on the results. Looking forward to your reply
There are doubts about the CNN features of the image dataset mentioned in your article. Are you using a base model (such as torchvision.models.resnet18(pretain=True) or others) to extract features or a custom CNN module? In my experiment, I found that different feature extractions still have a great impact on the results. Looking forward to your reply