Quick Links

• Homepage
• Google Scholar
• LinkedIn

---

Projects

Deep Metric Learning Baselines

Implementations of various Deep Metric Learning standards in a fashion that allows for easy extension and inclusion of novel methods. Provides standard architectures (ResNet50, GoogLeNet) with respective evaluation protocols, triplet sampling methods and loss functions.
See this link for more info.

Liver Lesion Segmentation for LiTS using 2D UNets

This repo is build around a variable UNet scaffold for different 2D architectures. Training can be performed using a multitude of loss functions (dice loss, pixel- or class-weighted cross-entropy loss, combined loss-functions…).
See this link for more info.