This notebook aims to make a Convolutional Neural Network (CNN) model for Kanji Classification using the Triplet Loss.
The goal is to create embeddings of Kanji drawings, which can then be compared with reference images using cosine similarity to perform classification.

You can install the dependencies required for this notebook from the requirements.txt file.

Warning : I can't provide the datasets used for this project here. More information about them is available directly in the notebook.