Hero Image
Autoencoder

Creation of pre-trained autoencoder to learn the initial condensed representation of unlabeled datasets. This architecture consists of 3 parts: Encoder: Compresses the input data from the train-validation-test set into a coded representation which is typically smaller by several orders of magnitude than the input data. Latent Space: This space contains the compressed knowledge representations and is thus the most crucial part of the network. Decoder: A module that helps the network to “decompress” the knowledge representations and reconstruct the data from their coded form.

January 10, 2022 Read
Hero Image
Skin Cancer Detection using semantic segmentation

Skin Cancer Detection Tool README.md Overview The objective of this project is to build a Skin Cancer Detection Tool. The tool that we are creating is a segmentation model of spots (moles, melanomas, etc…) on microscopic images of the skin. To create this tool we will have to train a semantic segmentation AI model. The data that we use for that training is from The International Skin Imaging Collaboration. File Descriptions: data.

August 10, 2021 Read