There is a newer version of the record available.

Published September 7, 2023 | Version v1
Other Open

Pre-trained μSplit

Creators

  • 1. Human Technopole
  • 2. University of Birmingham
  • 3. University of Pavia
  • 4. Univerisity of Pavia

Description

It contains pre-trained Deep-LC and Lean-LC variants of μSplit meta-architecture. Modified Hierarchical Variational Autoencoder architecture is used. Refer to the paper "μSplit: efficient image decomposition for microscopy data", https://arxiv.org/abs/2211.12872 for more details. The associated code hosted at https://github.com/juglab/uSplit should be used. Evaluation.ipynb computes the mean and stdev from the data and uses that to normalize the input before feeding it into the model.

Files

uSplit_trained_models.zip

Files (463.2 MB)

Name Size Download all
md5:18eefa6b4c15902d4af81ecf21e3e609
1.1 kB Download
md5:921afe71fa7de15c90c13345bc346f45
463.2 MB Preview Download