Building a shape model
To build a shape model from the data, create a parameter file in the models director, for example: build_shape_model.params.
From within the models directory, run:
msm_build_shape_model -p build_shape_model.params
this will create a binary file, shape_model.bfs containing a msm_shape_model.
Displaying the modes of the model (command line)
You can use the tool msm_draw_shape_modes to create one or more EPS files showing the modes.
First create a parameter file, for example draw_shape_modes.params.
Then call:
msm_draw_shape_modes -p draw_shape_modes.params
This will generate one file per mode, showing the mean shape and the effect of varying each parameter by +/-3SDs.
If you change the parameter make_movie to true, it will generate a sequence of movie frames, which can be collected and displayed with your favourite animation tool.
Displaying the modes of the model (GUI)
Use the GUI tool, qmsm_shape_mode_viewer:
qmsm_shape_mode_viewer -s shape_model.bfs -c face_68pts.crvs
Building a single ASM
To build a single ASM, create a parameter file in the models directory, for example: build_asm1.params.
From within the models directory, run:
masm_build_asm -p build_asm1.params
This builds a binary file, face1.masm, containing an ASM object.
Building a series of ASMs
First build a second ASM.
Create a second parameter file, for example build_asm2.params.
Run:
masm_build_asm -p build_asm2.params
Set up a new parameter file defining which single ASMs to combine.
For instance, make_asm_series.params:
// Directory for the models
model_dir: ./
// List of built models
model_files: {
face1.masm
face2.masm
}
Run:
masm_make_asm_series -p make_asm_series.params -o face_series.masms
Searching new images using a GUI
You can use the qasm_markup_tool to search new images with the model you have just built.
Testing performance of ASMs
Use the masm_test_search tool.
Set up a parameter file, for instance test_search.params.
The tool allows you to initialise the shape model points with examples from one directory (initial_points_dir), run the
series of ASMs, then compare the result with points defined in a second directory (points_dir).
If initial_points_dir=points_dir then this just gives an idea of whether the model converges given the correct starting point.
Run:
masm_test_search -p test_search.params
UoMASM Index