Skip to content

Latest commit

 

History

History
36 lines (32 loc) · 1.07 KB

File metadata and controls

36 lines (32 loc) · 1.07 KB

Tasks

Priority

  • add wandb to training setup for RAITE - refactor training code to run w wandb ******
    • [DONE] succesful training with wandb
    • clean wandb workflow - ensure useful metric are being tracked
    • save models to model repo in wandb
  • CLI for training
  • CLI for inference/ visualization

Not Priority

  • [DONE] CLI for help page
  • CLI for evaluate
    • [DONE] eval on json only
    • [DONE] evalute on dataset vs json options
    • [DONE] add CLI flexibilty for diff image sizes
    • [DONE] add CLI for different mapping + explanation
    • have single eval produce results file
  • CLI for multi-model evaluation
  • determine tp_fp_fn function diff
  • add type hints
  • add batch training
  • modify raitedataset object call to be flexible for other input parameters
  • run pylint on all files and fix warnings (pylint)
  • run formatter on all files (black)
  • debug final_ap
  • clean up comments
  • clean up readme

Tests for evaluation:

  • different image sizes 400x400, 800x800, 400x800
  • different mappings {0:0, 0:1}?

Later Priority

  • connect with isaacsim