Deep Learning using Geometric Features
The Active Segmentation platform for ImageJ (ASP/IJ) was developed in the scope of GSOC 2016 – 2018. The plugin provides a general-purpose environment that allows biologists and other domain experts to use transparently state-of-the-art techniques in machine learning to achieve excellent image segmentation. ImageJ 3 is a public domain Java image processing program extensively used in life and material sciences. The program was designed with an open architecture that provides extensibility via plugins.
The project idea: The existing machine learning model of Active Segmentation is based on the Weka library. However, this is limited to traditional machine learning approaches. The objective of the project will be to incorporate deep-learning functionality into the platform. Deep neural nets are capable of record-breaking accuracy. The candidate will explore existing implementations, for example Deeplearning4j 2 or Neuroph. Next, the candidate will select a library to be incorporated into ASP/IJ and will provide a reference implementation.
Tasks
● Fix existing issues and bugs
● Add-on to the user interface for trajectory display
● Provide a reference implementation
Minimal set of deliverables
● Requirement specification – Prepared by the candidate after understanding the functionality.
● System Design – Detailed plan for the development of the plugin and test cases.
● Implementation and testing – Details of implementation and testing of the platform.
Desired skills: Java, machine learning
Mentors: Dimiter Prodanov ([email protected] ), INCF Belgian Node; (backup) Sumit Vohra, ZIB, Berlin, Germany
References
