A precise segmentation model trained on the ISIC2016 and 2017 datasets. Throws an accuracy of 98.06% and a Jaccard Index of 90.86. Based on the U-Net architecture with a DenseNet201 backbone.
- Downloads last month
- -
Model tree for DevBhuyan/Skin-Lesion-Segmentation
Base model
Sadiksmart0/unetDatasets used to train DevBhuyan/Skin-Lesion-Segmentation
Evaluation results
- accuracy on isic2016self-reported98.040
- precision on isic2016self-reported97.090
- IoU (jaccard index) on isic2016self-reported90.860
- F1-score (dice coefficient) on isic2016self-reported94.780
- accuracy on isic2017self-reported93.060
- precision on isic2017self-reported98.630
- IoU (jaccard index) on isic2017self-reported89.970
- F1-score (dice coefficient) on isic2017self-reported94.350
# !pip install -U keras tensorflow huggingface_hub # Keras needs TensorFlow installed to read "hf://" paths, so the tensorflow backend is selected here; # "jax" and "torch" also work for computation once TensorFlow is installed. import os os.environ["KERAS_BACKEND"] = "tensorflow" import keras model = keras.saving.load_model("hf://DevBhuyan/Skin-Lesion-Segmentation")