Uploaded image for project: 'SAFe Program'
  1. SAFe Program
  2. SP-3511

SRC workflow example: Source Classification (CNN)

Change Owns to Parent OfsSet start and due date...
    XporterXMLWordPrintable

Details

    • SRCnet
    • Hide

      SRCs will provide pre-built scripts for standard workflows that users expect to run. We aim to implement some of these, and test their effectiveness with regards to hardware limitations and data storage schemes. 

      This will be a small task that looks at source classification using a Convolutional Neural Network (CNN). It will use pre-prepared data from the NVSS and FIRST radio surveys, and run a CNN on a GPU to classify them. To simplify this task, it will not include data cleaning or preparation, and use the pre-prepared dataset MiraBest. This will be the first GPU based task, which will be important for testing GPU compute infrastructure. 

      Show
      SRCs will provide pre-built scripts for standard workflows that users expect to run. We aim to implement some of these, and test their effectiveness with regards to hardware limitations and data storage schemes.  This will be a small task that looks at source classification using a Convolutional Neural Network (CNN). It will use pre-prepared data from the NVSS and FIRST radio surveys, and run a CNN on a GPU to classify them. To simplify this task, it will not include data cleaning or preparation, and use the pre-prepared dataset MiraBest. This will be the first GPU based task, which will be important for testing GPU compute infrastructure. 
    • Hide

      AC1: Reproduce a radio galaxy classifier using a CNN using data from the NVSS and FIRST radio surveys using the MiraBest dataset.

      AC2: Containerise the task and deploy it on the SRC Workloads gitlab repo. 

      AC3: Record a demo (not necessary until the GPU part is done)

      Show
      AC1: Reproduce a radio galaxy classifier using a CNN using data from the NVSS and FIRST radio surveys using the MiraBest dataset. AC2: Containerise the task and deploy it on the SRC Workloads gitlab repo.  AC3: Record a demo (not necessary until the GPU part is done)
    • 1
    • 1
    • 0
    • Team_MAGENTA
    • Sprint 5
    • Hide

      Code: https://gitlab.com/ska-telescope/src/src-workloads/-/tree/master/tasks/cnn-classifier-mirabest?ref_type=heads 

      Demo: https://drive.google.com/file/d/1qXq7wiyYUtAcxECapfItS6qhvUgZ-tn0/view?usp=drive_link 

      Show
      Code: https://gitlab.com/ska-telescope/src/src-workloads/-/tree/master/tasks/cnn-classifier-mirabest?ref_type=heads   Demo: https://drive.google.com/file/d/1qXq7wiyYUtAcxECapfItS6qhvUgZ-tn0/view?usp=drive_link  
    • 20.6
    • Stories Completed, Outcomes Reviewed, Demonstrated, Satisfies Acceptance Criteria, Accepted by FO

    • PI24 - UNCOVERED

    • PI20-PB

    Description

      This workflow uses a CNN to classify images of radio galaxies from the FIRST and NVSS surveys.

      The published workflow: https://gitlab.com/ska-telescope/src/src-workloads/-/tree/master/tasks/cnn-classifier-mirabest?ref_type=heads

      Video demo: https://drive.google.com/file/d/1qXq7wiyYUtAcxECapfItS6qhvUgZ-tn0/view?usp=drive_link

      Stretch: edit the workflow to use GPUs.

      Attachments

        Issue Links

          relates to

          Feature - A Feature is a service that fulfills a stakeholder need. Each feature includes a benefit hypothesis and acceptance criteria, and is sized or split as necessary to be delivered by a single Agile Release Train (ART) in a Program Increment (PI). SP-3496 Consolidate naming and structure of SRC example workflow repository

          • Must have - Items labelled as "Must have" are critical to the current delivery timebox in order for it to be a success.
          • Done
          mentioned in

          Page Loading...

          Page Loading...

          Structure

            Activity

              People

                P.Llopis Llopis, Pablo
                r.bolton Bolton, Rosie
                Votes:
                0 Vote for this issue
                Watchers:
                2 Start watching this issue

                Feature Progress

                  Story Point Burn-up: (100.00%)

                  Feature Estimate: 1.0

                  IssuesStory Points
                  To Do00.0
                  In Progress   00.0
                  Complete57.5
                  Total57.5

                  Dates

                    Created:
                    Updated:
                    Resolved:

                    Structure Helper Panel