Diamantis99/segformer_mit_b5

Segformer Model Card

Table of Contents:

• Load trained model
• Model init parameters
• Model metrics
• Dataset

Load trained model

import segmentation_models_pytorch as smp

model = smp.from_pretrained("<save-directory-or-this-repo>")

Model init parameters

model_init_params = {
    "encoder_name": "mit_b5",
    "encoder_depth": 5,
    "encoder_weights": "imagenet",
    "decoder_segmentation_channels": 256,
    "in_channels": 3,
    "classes": 1,
    "activation": None,
    "upsampling": 4,
    "aux_params": None
}

Model metrics

[
    {
        "test_per_image_iou": 0.5847598910331726,
        "test_dataset_iou": 0.7008192539215088,
        "test_per_image_accuracy": 0.8566379547119141,
        "test_dataset_accuracy": 0.8566379547119141
    }
]

Dataset

Dataset name: VIP

More Information

• Library: https://github.com/qubvel/segmentation_models.pytorch
• Docs: https://smp.readthedocs.io/en/latest/

This model has been pushed to the Hub using the PytorchModelHubMixin