v1.6.3rc3

Update VERSION

v1.6.3rc2

Improved performance of ensemble

bugfix

• changed tempdir to output directory

Added --gpu flag and now install IMOD

• Added IMOD to list of software installed by cloudformation template. Issue #52

• Added --gpu option to runprediction.sh Issue #51

• Added DatasetURL parameter to CloudFormation Template which if set to URL downloads file specified.

bug fixes and --gpu added to runtraining.sh

• Fixed --1fmonly flag causing error when used passed to runtraining.sh bug. Issue #47

• Added --gpu option to runtraining.sh. Issue #48

• Fixed EnsemblePredictions.m fails when called by runprediction.sh if there is only 1 model. Issue #49

Parallel segmentation and prediction

• Enable parallelism in prediction #33

• Switch runprediction.sh pipeline to augment data on the fly #36

• Enable parallel training #38

Addition of tunable parameters for training

Fixed following issues

• Update default number of iterations in runtraining.sh to 30000. Issue #31

• Update runtraining.sh with custom flags to allow caller to adjust learning rate etc. Issue #30

• Update prediction to run with less then 16 augmented variations of the data. Issue #32

• Copy over VERSION file to train and predict jobs to. Issue #35

Bug fix in merge

Fixed a bug where merging large datasets was failing

Updates to EnsemblePredictions.m

Updated EnsemblePredictions.m to ignore non .png files in #fm directories.

Added large pretrained mito dataset

v0.14.1

new release