The platform for ImageJ (ASP/IJ) was developed in the scope of GSOC 2016 - 2019. 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 and classification. ImageJ 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 active segmentation supports both traditional machine learning and deep learning approaches. However, the existing implementation supports a limited way to load the ground truth for learning i.e. region of interest (roi)file format. This forces users to convert their ground truth to specific format and which is not even suitable for multiple applications. The objective of the project will be to incorporate several ground truth formats into the platform e.g. image-based in which each pixel uniquely belongs to a particular class, partial ground truth format in which instead of whole image, several partial boxes in an image or stack are labeled, User need to customize metafile format in order to load such a partial ground truth.
On the similar note, the candidate will also allow several output formats to store segmentation and classification results. For example, output only the user selected region of interest or partial stack.
Tasks:
- Fix existing issues and bugs
- Get familiar with input/ out format i.e. image based, region of interest based, JSON.
- GUI implementation and integration
Minimal set of deliverables:
- Requirement specification - Prepared by the candidate after understanding the functionality.
- System Design - Detailed plan for development of the plugin and test cases.
- Implementation and testing - Details of implementation and testing of the platform.
Desired skills: Java, Machine learning
Mentors: Sumit Vohra, ZIB, Berlin, Germany; Dimiter Prodanov @dprodanov (dimiterpp@gmail.com ), INCF Belgian Node (backup)
References:
- ImageJ: https://imagej.nih.gov/
- Weka https://www.cs.waikato.ac.nz/ml/weka/
- Active Segmentation : https://github.com/sumit3203/ACTIVESEGMENTATION
- Deeplearning4J: https://deeplearning4j.org/