Update workshop materials based on feedback from April

[kmcdono2]

Using feedback from workshop (Data/Culture Sharepoint Documens --> LwM workshop reporting), these are things we will adapt for the June 2024 workshop.

• make section in documentation site about workshops. has pages: "how to organise a workshop", "workshop resources", "post workshop tips", "places to get maps for MapReader" #423
• make a document about how to organize MapReader workshop, e.g. #439
  • split people with PCs/Macs into different tables
  • organise pre-workshop zoom to trouble shoot install, depending on audience?
  • make video for installation (already a @rwood-97 plan) #421
  • clarify what prep to do in advance: read MapReader paper, programming historian lessons about CV/ML, python cheat sheet, juptyter notebooks, etc.
• make a "post workshop" page in documentation with tips for next steps #422
  • how to stay in touch
  • how to find new models and share yours
  • how to share datasets
• worked example and documentation with workflow for people who want to work with their own maps #440
• review notebooks and provide suggestions for how to frame "your turn" sections #420
• early on in workshop, have people look at felt map of railspace data to talk about nature of outputs, error, success, etc. so they understand the potential
• hyperlinks to relevant docs within notebooks
• explain training, validation sets
• explain epochs, loss and how this allows you to fine-tune your model
• talk about building footprint task
• encourage teamwork/propose shared annotation task
• comment each command in version of notebook that is for python beginners
• have different versions of notebooks: completely commented and filled in vs. bare bones
• more guidance about annotating, e.g. explain typical devisions around guideline development, importance of "none" label
• reduce CV/ML intro & model/train slides to what you need to know specifically for MapReader
• lunch 45 min, not 60?
• Finish working on "coding basics" page in documentation #441
• notebook for working with text spotting task output
• for free time: guidance for helping people do something in QGIS with their output file
• consider working on IIIF as input to make it easy for people to grab those (non-georef) maps

[rwood-97]

Also need to address:

• Add the device argument for GPU support on the ClassifierContainer() #411
• Update annotated_images.show_sample() example #406
• Broken URL in the Workshop_AprilMay2024_part2.ipynb notebook #408
• Invalid syntax error on the calculate_add_metrics cell #409
• Displaying scores per class #410

[rwood-97]

Before starting notebooks:

• explain the comment and how to fill in commented sections
• explain passing variables to functions

Suggestion to perhaps include in part2 notebook:
After the inference on the first GLASGOW sheet, to run inference on another map sheet (or include this as an exercise - it just involves changing the sheet ID in a couple of places). One participant was really pleased to see how good the results were on other sheets.

[rwood-97]

@katie can I close this ticket since all the other main unfinished points have their own issues now