https://hamptonjc.github.io/skin-lesion-classifier-web-app/
This web-app uses a Mobilenet neural network to classify images of skin lesions (i.e. moles).
The network was developed in Python with Tensorflow.
The training process is available in a Colab notebook -> link
The network started from a pretrained Mobilenet. The architecture was modified to fit the problem. The network was trained on the HAM10000 dataset ->link
The network was converted into a TensorflowJS model. The web-app was developed with Angular & Angular Material.