Our CAST can also be trained with fully-supervised learning. In the paper, we fine-tune pre-trained models on ADE20K and Pascal Context for semantic segmentation. We pre-train models for fully-supervised classification with DeiT framework.
We provide the bashscripts for running fully-supervised experiments. By default, we use CAST-S
. You can use larger models, e.g. CAST-B
by replacing --model cast_small
with --model cast_base
in the bashscripts.
- fully-supervised learning of CAST on ImageNet-1K:
> bash scripts/deit/train_imagenet1k_cast.sh