Figure 1. The schematic of the ML Ecosystem of Filament Detection. The deliverable Data Products (DP) and Software Products (SP) are labeled below each card. The six cards from left to right correspond to DP1, DP2, SP1, SP2, SP3, and SP4 in Fig 1.
Figure 2. The timeline of the project, broken down into the 6 deliverables. Data Products (DP1-2) and Software Products (SP1-4) are the same as those in Fig 1.
Deliverables. A stand-alone software providing the following key features:
Uniform sampling of H-Alpha images from the GONG network based on sample size or sampling cadence
Conversion of FITS images to JPEG format with a desirable quality for the annotation process
Automatic detection of corrupt images
Dynamic distribution of images to groups of annotators
Structuring data for quality control of the annotation process
Deliverables. N/A
A team of trained annotators will be working on annotation of filaments while our developers will continuously monitor the quality of the annotations providing feedback loop to the annotation team.
Deliverables. An interactive web app to help the annotators explore the evolution of solar filaments on H-Alpha images.:
The publicly accessible web app
The code base of the website with documentation
Deliverables. N/A
Development of a Database System (PostGIS) to store the annotated data
Analysis of the so-far created annotations and highlighting the possible issues
Testing an object segmentation algorithm on our data (to find out the need for further cleanups.)
Deliverables. Updates on the web app
Connecting the PostGIS DB, API, and Web app into a modular system.
Adding new features to the web app for showing the annotations on the observations.
The manually-annotated dataset is releaed.