2017-12-04 - Fiji + KNIP hackathon

Revision as of 08:50, 15 December 2017 by Pbainar (talk | contribs) (Adding conclusion of parallelization technical discussions)

From Monday, December 4, 2017 through Friday, December 15, 2017, the Max Planck Institute of Molecular Cell Biology and Genetics hosts ~50 developers at the Center for Systems Biology in Dresden, Germany for a hackathon to develop ImageJ2 and Fiji core infrastructure and plugins.


Voluntary hackathon calendar


Hackathon google doc


Hackathon on Twitter (#hackdd17)


Technical Discussions

N5, not HDF5

For more info on N5, check out the github repository here.

  • feels like HDF5, but stores chunks(blocks) in separate files in the file system.
  • is a Java library, but Constantin Pape already wrote a C++/python version of it: z5 (also matches "zarr" library), https://github.com/constantinpape/z5
  • attributes are stored in an additional JSON file
  • Discussion: should we define standard now as to how data should be stored in there to prevent an emergence of a zoo of different flavors as there is for HDF5?
    • how to do time series where each timestep / angle could have different image size
    • if we want a general "N5 viewer" for images, we'd have to add calibration data
    • put this information around the N5 dataset, because it behaves more like a dataset within an H5 file.
    • Perhaps make it versioned? Because a duck is not always a duck...
  • why another file format?
    • parallel writes (awesome for clusters with shared filesystem)
    • there is a special type for label blocks
    • blocks can have a halo
    • the block grid does not need to be filled dense, some blocks could be missing
    • couldn't this just be another flavor of HDF5?
  • are parallel writes to the same block prevented by some kind of locks?
  • the HDF5 team should be included in the discussions to learn from their mistakes - there is lots of information on parallel writing of HDF5 files out there
  • Try to write a N5 dataset into a FUSE filesystem/file??? Could this be a work-around for the many-small-files issue?


  • There is a fork of BigDataViewer for JavaFx. [1]
  • BigDataViewer will be splitted into a UI independent part, and the Swing UI. This will make it possible to merge the JavaFx and Swing Version of BigDataViewer .

We discussed opportunities to improvement for the Bdv design:

  • Slim the core of bdv, make it less smart and more predictable.
  • Bdv handle for different UI toolkits,
  • Use property pattern for settings like active source ...
  • Actively add/remove views
  • Update views (force cache-dropping) - for BDV add/remove views
  • Access to color settings, overlap rendering, …
  • Allow Grouping of views (externally is ok)
  • Consider Accumulator / Composite that works with type other than ARGBType
  • Make it easy to disable or replace Dialogs
  • Make it easy to remove an overlay
  • BdvVistools show function for converters/volatile

Matrix and Vector libraries

ImageJ/Fiji parallelization

  • initial focus on embarassingly parallel data
  • specific parallel (cluster) implementation will be abstracted
  • from the system architecture point of view, the design can be inspired by the search bar feature implementation
  • started use case definition

Hackathon Progress

ilastik - (Dominik Kutra, Carsten Haubold)

  • developed an ImageJ2 plugin that allows to stream raw data and predictions from the work in progress ilastik processing backend: https://github.com/ilastik/org.ilastik.bdvsource
  • discussed with Philipp Hanslovsky about the benefits of using the N5 format for communication
    • the label block format together with the fact that datasets can be stored sparse could be a great foundation for label storage / communication
    • played around with Constantin Pape's C++/Python N5 reader as available format on the ilastik server
    • noticed that the only thing needed to serve a N5 dataset on the web is to serve the directory contents. Philipp wrote a N5 REST reader for BigCAT to talk to the ilastik server that way
  • worked on some caching problems in the ilastik server
  • discussed possible interfaces between Matthias Arzt's LabKit and ilastik-backend for interactive segmentation in ImageJ/BigDataViewer with an ilastik classifier backend

SciView - (Kyle Harrington, Ulrik Günther, Curtis Rueden)

  • set up automatic deployment of Fiji update site that works on all operating sustems
  • imglib2 images as textures on 3D objects
  • imagej2-style logging
  • public availability of SciView plugin

ImageJ server - (Curtis Rueden, Petr Bainar)

  • discussed its design with respect to the ImageJ/Fiji architecture as well as its future use cases
  • identified the most imminent issues, raised them on GitHub and brainstormed possible solutions
  • worked on a new JavaScript client with enhanced UI, making use of the Angular framework
  • demoed the new client to the community
  • encouraged members of the community to use annotations specifying whether a SciJava plugin can run headless or not