Project information

  • Category: Non-supervised machine learning
  • Client: Universidad de Buenos Aires
  • Project date: september 2023
  • Project URL: Link to project

Description

In this project we processed images with a Convolutional Neural Network called VGG16, keeping the features obtained from the layer before the last one. Then we used different clustering algorithms (K-means, K-medoids, DBSCAN) to cluster the images based on the features obtained and compared the results with Van-Dongen and Rand Index. In order to visualize the results, we used algorithms for dimensional reduction (PCA and T-SNE). Finally we applied some algorithms used for object detection in images.