| # Utility functions used in preprocessing steps | |
| import glob | |
| import os | |
| from torchvision.transforms import Resize, ToPILImage, ToTensor | |
| import networkx as nx | |
| # Returns all the files paths with a specific extension inside a requested root directory | |
| def get_paths(rootdir, ext="png"): | |
| paths = [] | |
| for path in glob.glob(f'{rootdir}/*/**/*.'+ext, recursive=True): | |
| paths.append(path) | |
| return paths | |
| # Cluster the images generating a graph of connected components | |
| def _generate_connected_components(similarities, similarity_threshold=0.80): | |
| graph = nx.Graph() | |
| for i in range(len(similarities)): | |
| for j in range(len(similarities)): | |
| if i != j and similarities[i, j] > similarity_threshold: | |
| graph.add_edge(i, j) | |
| components_list = [] | |
| for component in nx.connected_components(graph): | |
| components_list.append(list(component)) | |
| graph.clear() | |
| graph = None | |
| return components_list | |
| # Method used to preprocess the image before features extraction in clustering step | |
| def preprocess_images(img, shape=[128, 128]): | |
| img = Resize(shape)(img) | |
| return img | |