Description of several approaches to the problem of image classification, for example, the definition of clothing attributes. The dataset uses images from google images. Each image is assigned 1 label from the set {blue, white, grey, black} and one label out of many {v neck, roundneck}.
- multihead.ipynb - is an example of using multi-head neural net to predict clothes attributes. Visualization of predictions using cad-cam.
- Siamese_mnist.ipynb - is a notebook showing approaches to Siamese networks on a simple task(mnist).
- Siamese_clothes.ipynb - is a notebook that solves the problem of classifying an individual attribute of clothing using Siamese networks.
- master_siamese.ipynb - shows an example of how to use the papermill library to turn a ipynb file into an 'executable script'. The resulting notebooks in the executed_nb folder.