Monet-or-Manet
This repository contains the source code for the Beyond Image Classification: Leveraging CNNs and Clustering Methods to Provide Informed Artistic Recommendations project. The dataset for the project can be downloaded from wikiart. The dataset was generated by the WikiArt Crawler
Project Details
The three folders in this repository are Analysis, Models, and Recommendation Engine.
Analysis
The analysis Folder contains the code for the analysis of the dataset. The three python notebooks plot the training and testing losses for pre-trained and predefined models over the 10%, 50%, and 95% subsets of the data. The analysis csvs are generated from the notebooks in the models folder.
Models
Python file in the Models folders contains all the models listed in the paper. The models are tested on the 10%, 50%, and 95% subsets of the data. The file, in combination with the model_utils file, was used to generate the analysis csvs.
Recommendation Engine
Recommendation Engine jupyter notebook utilizes the layer slicing and the model_utils file to generate the recommendations as discussed in the paper.