sgugger/resnet50d

The ResNet architecture was introduced in this paper and is adapted with the ResNet-D trick from this paper . The training images in the dataset are usually photos clearly representing one of the 1,000 labels . The model has a top1-accuracy of 80.53% and a top-5 accuracy of 95.16% in the evaluation set of ImageNet images. The model will probably not generalize well on drawings or images containing multiple objects with different labels. It will work better on images picturing scenes from the US (45.4%) and Great Britain (7.6%) As such the model or. models created by fine-tuning this model will . work better,