Images Classification Recipe

Images Classification

Convolutional Neural Networks (CNN) and Transfer Learning for Heritage Images Classification

This recipe exposes an image classification scenario, aiming to deduce the technique or genre of images (picture, drawing, map...) using a Convolutional Neural Networks trained model (supervised approach with transfer learning).

This recipe may have various library use cases, particularly for cataloguing and information retrieval systems.

This recipe is based on BnF materials

Goals

This recipes includes a basic introduction to neural networks and deep learning (formal neuron model, neural networks, convolutional neural networks, transfer learning) and a hands-on session:

1. Creation of an images dataset for training
2. Training of a classification model with commercial APIs or open source IA frameworks
3. Application of the model to the heritage images to be classified

The IIIF standard API is leveraged to extract images from digital repositories, but raw files may also be used.

Educational resources

Introduction to the core technology used

For the theory, see the BnF github for a 45 mn introduction course (FR and EN versions).

Implementation notes

• prose description
• resources and skills needed
• the technical information needed to complete a sub-task of the recipe, or the whole recipe (Jupyter notebooks, detailed Git readme...)
• training datasets used ...

Other resources

• IBM Watson documentation
• Google AutoML documentation Convolutional neural networks:

• https://towardsdatascience.com/simple-introduction-to-convolutional-neural-networks-cdf8d3077bac (EN)
• https://www.supinfo.com/articles/single/8037-deep-learning-reseau-convolution (FR)

• Library of Congress Newspaper Navigator
• GallicaPix github

...