Guidance for Using the Pretrained Model
#4
by
Lahav222
- opened
Hello, I'm using your pre-trained ResNet50 model on CIFAR-100 dataset.
I have a directory with 10,000 test images(from the CIFAR-100 test set) in .jpg format and to apply the pre-trained model to each image, I'm using the following transform:
'transform = transforms.Compose([transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])])'
However, the accuracy(how much of the test images were classified correctly/10000) I obtained for the model is 0.61, which differs from the 0.809 on your page. I suspect I'm making a mistake but I'm not sure what it is. Any guidance would be appreciated.
Thanks,
Lahav.