Clothes RecSys with VAE inside

Dataset based on SSENSE website also known as FOTOS dataset was used. SSENSE.com is a popular fashion website, where user can create and upload outfit data. Dataset consists of 11,000 well-matched outfit composed by 20,384 fashion items. For now I will only use images that are related to looks consisting of tops-bottoms and bottoms-shoes. So, there are about 8500 outfits.

Input images had shape 680x680 and were resized to 128x128, latent space has 64 dimensions. Encoder has 5 convolution layers each with batch normalization and LeakyReLU activation function.