Date

Attendees

Discussion items

New version of tools catalog is still a focus to release


TimeItemWhoNotes


Shannon

Box Blog - releasing Friday

Release by NCSA tomorrow - Shannon will copy that to Brown Dog blog page - and retweet



Sandeep

ArcGIS docker

refactoring task will continue for another week

Emailing Mattias about images in Galaxy tool

something



Mark

Working on creating a Docker container for BDFiddle. Took a bit of extra time, as I work through the differences between running a single container that gets BDFiddle working, or making the BDFiddle container rely on other containers that would be started by a docker-compose.

Readme: is wrong - we do not need MySQL

Rob pointing to docker container used in Pecan

Fence is not the latest - should be easier



Deren




Luigi

Still need to reply to Louisville team - there are a couple extractors we need to create

They have code from MIT - identifies trees and grassy images - Dukyun Nam take it and create an extractor

How many pixels are in a picture - Mark Fredricksen can work on this

Share documentation with them - and then have a meeting Sandeep Puthanveetil Satheesan to show them how to use bdcli and bdfiddle



Rob

whole Geostack is running on dev / issue with Prod

working on logging in swarm for docker to have a way for log files to go to multiple places - container logs too

Elastic Search is on apache lucine - they have a module which is called XPack - was paid - did monitoring and analysis - as well as do log in

ES released everything except auth into Opensource - Amazon questioned actual opensourceness

Amazon will have an implementation that will allow auth soon that will allow log in - called Open Distro

Process that manages sending logs from docker to elastic search is in go - learning new skills



GregI do have a significant update. I've shared already that I've taken the punch card extractor project on, instead of mentoring students for that one. Today I can say that I've gotten a Python module and CLI up and running. It can handle backlit or frontlit card images, and either "landscape" or "portrait" orientations, and any kind of flipping (photos of the back of a card). This was pretty fun for someone relatively new to image processing. The current code is up and next step is to package it in an extractor and aggregate results at the dataset level. https://github.com/UMD-DCIC/punchcards 


Dukyun Nam

Joining Team - ML, Deep learning, clowder etc help - Meeting with Luigi tomorrow to spin up
SMU - with clowder and tensor flow - ways of getting data out of Clowder and into tensor flow and back

TERRA - is there a need to take the metadata associated with image that can be used to predict what an image is

Project - text in PDFs - extracting tables from PDFs - numbers in tables - identifying what value for a certain parameter or entry in table

To Dos - Tasks


  •