Edit model card

Intro

This is the model for our paper "Melanoma Detection using Adversarial Training and Deep Transfer Learning". Code is available here.

Model description

The model is trained on the ISIC 2016 Task 3 dataset. The architecture and algorithm is described in this paper.

Intended uses & limitations

You can use the raw model for melanoma detection from skin lesion images.

How to use

See Spaces demo. For more code examples, we refer to this GitHub deploy section.

Limitations and bias

The model is trained on a specific dataset with just over a thousand samples. It may or may not work for other kinds of skin lesion images. Further, there is no out-of-distribution detection method to filter out non skin lesion images. If you give an image of a dog, the model will still classify it as benign for malignant!

Training data

See dataset details.

Training procedure

See training details.

Evaluation results

For results in benchmarks, we refer to Figures 5, 6 and Table 1 of the original paper here.

Citation

@article{zunair2020melanoma,
  title={Melanoma detection using adversarial training and deep transfer learning},
  author={Zunair, Hasib and Hamza, A Ben},
  journal={Physics in Medicine \& Biology},
  year={2020},
  publisher={IOP Publishing}
}
Downloads last month
0
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Space using hasibzunair/melanet 1