search

uncbiag / registration

Image Registration
275 stars 32 forks source link

implementations? #4

Open Borda opened 4 years ago

Borda commented 4 years ago

Hello, I was wondering if you could also share code for samples/illustrations presented in this repo. It would be very useful... Thx :]

hbgtjxzbbx commented 4 years ago

Hi Borda Thanks for your interests. For the code, you can refer to https://github.com/uncbiag/mermaid/blob/master/docs/source/notes/rdmm_example.rst

Borda commented 4 years ago

that is an example, I was more thinking about providing the particular code you used for the animations, so it would be a great starting point...

hbgtjxzbbx commented 4 years ago

Sorry, the animation code is not added for now. I will update the code for animation later ( hopefully this week).
The script example_2d_synth.py can output the frame at any time point. To do that, you need first get the optimal momentum by running optimization once. Then, vary the "tTo" parameter ( the default is 1.0) to 0.1, 0.2 ....0.9, 1.0. Run the evaluate part multiple time and get pngs for each time point. We use online tools to generate the gif from pngs.

Borda commented 4 years ago

@hbgtjxzbbx any update here? 🐰

hbgtjxzbbx commented 2 years ago

@hbgtjxzbbx any update here? rabbit

Hey Borda I might mistake your point. The link mentioned above (https://github.com/uncbiag/mermaid/blob/master/docs/source/notes/rdmm_example.rst) includes toy data generation and supports several fluid-based registration approaches including in this repository, i.e., vsvf, lddmm, rdmm.

The optional settings in example_2d_synth.py are as follows:

Registration demo for 2d synthetic data

optional arguments:
  -h, --help            show this help message and exit
  --expr_name EXPR_NAME
                        the name of the experiment
  --data_task_path DATA_TASK_PATH
                        the path of data task folder
  --model_name MODEL_NAME
                        non-parametric method, vsvf/lddmm/rdmm are currently
                        supported in this demo
  --use_predefined_weight_in_rdmm
                        this flag is only for RDMM model, if set true, the
                        predefined regularizer mask will be loaded and only
                        the momentum will be optimized; if set false, both
                        weight and momenutm will be jointly optimized
  --mermaid_setting_path MERMAID_SETTING_PATH
                        path of mermaid setting json
mlcJoin commented 2 years ago

The email has been received   thanks.