Design & spec out blind labeling feature

[mihow]

• What does UKCEH / AMBER project need
• What do Aditya / Yuyan desire for sampling
• New collections feature for crops instead of source images.
• What does ML need to improve models
• What does research paper need if needed (strictness about bias, etc)
  • Is it okay to use our detector and binary classifier?
• What UI features do we need
• What backend features do we need
• Diagram of the entire feature
• Align on agreements & timelines

MVP: register new ML pipeline with only detector & moth/non-moth classifier

[joseepoirier]

Resumed conversation with Jenna. We will meet with them soon, ideally before 26 Sept.

[mihow]

I met with Jenna and walked through the platform to see if it could meet their current needs. There are a few things that may need to be done manually by Michael / me on the backend (make a collection for specifically 1-2 deployments). But not a ton other than getting the production environment ready. A new walkthrough video for the annotation project could be very helpful - especially showing how the current filters work. There are a couple easy UI changes that could help.

Notes & timeline here https://docs.google.com/document/d/1GsrCnxpbDGKmEWktEfQO3F-3ppnrp6dMu8QPV6UQDP0/edit#heading=h.iganhh7knnn