metadata
library_name: keras-hub
This is a SegFormer model uploaded using the KerasHub library and can be used with JAX, TensorFlow, and PyTorch backends.
This model is related to a ImageSegmenter task.
Model config:
- name: seg_former_backbone
- trainable: True
- projection_filters: 768
- image_encoder: {'module': 'keras_hub.src.models.mit.mit_backbone', 'class_name': 'MiTBackbone', 'config': {'name': 'mi_t_backbone', 'trainable': True, 'layerwise_depths': [3, 4, 18, 3], 'hidden_dims': [64, 128, 320, 512], 'image_shape': [512, 512, 3], 'num_layers': 4, 'layerwise_num_heads': [1, 2, 5, 8], 'layerwise_sr_ratios': [8, 4, 2, 1], 'max_drop_path_rate': 0.1, 'layerwise_patch_sizes': [7, 3, 3, 3], 'layerwise_strides': [4, 2, 2, 2]}, 'registered_name': 'keras_hub>MiTBackbone'}
This model card has been generated automatically and should be completed by the model author. See Model Cards documentation for more information.