This model is a fine-tuned version of
google/vit-base-patch16-384 on the imagefolder dataset.
It achieves higher accurate than 224 model.
A custom dataset about 28k images, if you need to improve your domain's accurate, you can contribute the dataset to me.
