Adding skeletonization notebooks to the repo. #59
Conversation
| "metadata": {}, | ||
| "outputs": [], | ||
| "source": [ | ||
| "# mesh_path = \"/scratch/testing/864691135014681590.obj\"\n", |
There was a problem hiding this comment.
No hard coded paths please.
Use the obi-notebook widget to find and download meshes.
| "source": [ | ||
| "environment = \"staging\"\n", | ||
| "token = get_token(environment=environment, auth_mode=\"daf\")\n", | ||
| "project_context = get_projects.get_projects(token, env=\"staging\")" |
There was a problem hiding this comment.
Use env=environment instead of env="staging"
| "id": "ed65fc07", | ||
| "metadata": {}, | ||
| "source": [ | ||
| "## Display EMCellMesh table" |
There was a problem hiding this comment.
Please add some instructions here. Explain what is being displayed and what a user is expected to do. And how they can understand what each mesh means.
| " \"description\": \"Reconstructed morphology from an EM surface mesh\"\n", | ||
| "}\n", | ||
| "\n", | ||
| "skeletonization_params = {\n", |
There was a problem hiding this comment.
For the skeletonization params:
I suggest using ipywidgets (maybe Slider, Checkbox) for the selection.
Not just because it is fancier, but also because this allows us to set a "reasonable" interval for the resolution parameters. Right now, I would not know what range of resolutions makes sense at all.
Also, with some callback functions the behavior can be such that the spine resolution is always better than the other resolution.
| " output_morphology_id = UUID(job.get('output').get('morphology').get('id'))\n", | ||
| " break\n", | ||
| " elif status == 'failed':\n", | ||
| " print(json.dumps(job, indent=2))\n", |
There was a problem hiding this comment.
This can be formatted better than json.dumps
I have added two notebooks for the skeletonization functionality. The first one is quite simple and uses a single function to run the skeletonization pipeline end-to-end. The utilities notebook provides a step-by-step execution with several visualization and visual validation capabilities.