Next steps plugin implementation

[constantinpape]

• Handle resolution and rescaling correctly.
  • Read resolution from input file if possible (for mrc / rec).
  • Otherwise use optional input parameter in plugin.
  • Determine correct scaling based on average resolution model was trained on. (Need to aggregate this information).
  • Use the resolution in the distance measurement plugin.
• Upload all models and connect them in the plugin.
  • I will put them in a folder on our drive.
  • Aggregate the training information, including average resolution (see prev. point).
  • Distinguish the different vesicle models: vesicles, vesicles_2d, vesicles_cryo
• Improve handling of tiling:
  • Use get_default_tiling to determine the tiling values in the plugin.
  • Update the tiling for the 2D model to be at least 512 x 512.
• Use napari progressbars instead of tqdm to display activity in napari. This is simple by replacing import tqdm with

try:
    from napari.utils import progress as tqdm
except ImportError:
    from tqdm import tqdm

[lufre1]

to use the napari progress bar we need to adapt https://github.com/constantinpape/torch-em/blob/4f7b180af8177da9f1c1c5b09b282f7c270a1249/torch_em/util/prediction.py#L7 in torch-em

[constantinpape]

to use the napari progress bar we need to adapt https://github.com/constantinpape/torch-em/blob/4f7b180af8177da9f1c1c5b09b282f7c270a1249/torch_em/util/prediction.py#L7 in torch-em

Can you make a PR in torch_em for this?