/SwinTransformer

Primary LanguagePython

SwinTransformer

SWIN Transformer in TensorFlow 2.x

Size free for inputs to support tasks like Image Segmentation

Offical Pytorch repo.

Weights

All weights below are fully tested.

Name links
Swin Large 384 weiyun
Swin Base 384 weiyun
Swin Tiny 224 weiyun

Usage

def load_h5_weight (model, path, skip_mismatch = False):

    with h5py.File(path, 'r') as f:
        if 'layer_names' not in f.attrs and 'model_weights' in f:
            f = f['model_weights']
        
        layers = get_all_layers(model)
        
        load_weights_from_hdf5_group_by_name(f, model.layers, skip_mismatch = skip_mismatch)