High-dimensional clustering using deep neural networks

Software / Service definition

High dimensional Clustering using deep neural networks.

Usage in COVID19

The module can enable researchers to conduct clustering on CT or x-ray images so that similar cases are grouped together. Network weights can be finetuned by the available ground truth or trained from scratch. Alternatively, the provided models can be used.

  • Pretrained models on image datasets COIL20,COIL100, MNIST, USPS, YTFaces, FashionMNIST
  • Networks: ResNet18, ResNet34
  • Docker Implementation for train & inference

Access

Code via repository, models/data via link

TRL

7

Support point of contact

Petros Daras - daras@iti.gr

Previous application scenario

Image Clustering, Indices/Scores clustering (H2020 NADINE project)

Licence

Free until end of pandemic as per WHO declaration

Resources