def run_code(num_points, ratio, sigma, smallest_number, name=''):
original_gaussian_cac = CACSegmenter(OriginalGaussianCAC)
new_parameters = {
'num_points': [num_points],
'ratio': [ratio],
}
original_gaussian_cac.parameters = new_parameters
original_gaussian_cac.sigma = sigma
parameter_list = original_gaussian_cac.get_parameters()
dataset = original_gaussian_cac.load_dataset('AlpertGBB07_input_subtest.txt')
results_folder = 'segment_alpert_old_gaussian/' + original_gaussian_cac.CAC.__name__ + name + '/'
print results_folder
original_gaussian_cac.test_model(dataset, parameter_list[0], results_folder, plot_evolution=False)
def run_code(num_points, ratio, sigma, smallest_number, name=''):
original_gaussian = CACSegmenter(MultivariateGaussianCAC)
new_parameters = {
'num_points': [num_points],
'ratio': [ratio],
'smallest_number': [smallest_number]
}
original_gaussian.parameters = new_parameters
original_gaussian.sigma = sigma
parameter_list = original_gaussian.get_parameters()
dataset = original_gaussian.load_dataset('AlpertGBB07_input_subtest.txt')
results_folder = 'segment_alpert_multivariat_gauss/' + original_gaussian.CAC.__name__ + name + '/'
print results_folder
original_gaussian.test_model(dataset,
parameter_list[0],
results_folder,
plot_evolution=False)
current_cage_obj,
alpha,
grad_k,
color=[0., 0., 255.]):
utils.plotContourOnImage(contour_coord,
self.image_obj.image,
points=current_cage_obj.cage,
color=color,
points2=current_cage_obj.cage -
alpha * 10 * grad_k)
if __name__ == '__main__':
input_filename = 'AlpertGBB07_input_subtest.txt'
output_folder = 'seg_im/'
multi_mixture_gaussian_gray_cac = CACSegmenter(MultiMixtureGaussianCAC)
new_parameters = {
'num_points': [20],
'ratio': [1.1],
'smallest_number': [np.exp(-200)],
}
multi_mixture_gaussian_gray_cac.parameters = new_parameters
multi_mixture_gaussian_gray_cac.sigma = 0.25
parameter_list = multi_mixture_gaussian_gray_cac.get_parameters()
dataset = multi_mixture_gaussian_gray_cac.load_dataset(input_filename)
results_folder = output_folder + multi_mixture_gaussian_gray_cac.CAC.__name__ + '/'
multi_mixture_gaussian_gray_cac.test_model(dataset,
parameter_list[0],
results_folder,
plot_evolution=True)
image = self.image_obj.gray_image
image_gradient = np.array(np.gradient(image))
# Calculate Energy:
omega_1 = g_energies.grad_gauss_energy_per_region(omega1_coord, affine_omega_1_coord, image, image_gradient)
omega_2 = g_energies.grad_gauss_energy_per_region(omega2_coord, affine_omega_2_coord, image, image_gradient)
energy = omega_1 + omega_2
return energy
def _plotContourOnImage(self, contour_coord, cage_obj, alpha, grad_k, color=[0., 0., 255.]):
utils.plotContourOnImage(contour_coord, self.image_obj.gray_image, points=cage_obj.cage, color=color,
points2=cage_obj.cage - alpha * 10 * grad_k)
if __name__ == '__main__':
input_filename = 'synthetic_locality_input.txt'
output_folder = 'experiment1/'
original_gaussian_gray_cac = CACSegmenter(OriginalGaussianCAC)
parameter_list = original_gaussian_gray_cac.get_parameters()
new_parameters = {
'num_points': [12],
'ratio': [1.05],
}
original_gaussian_gray_cac.parameters = new_parameters
original_gaussian_gray_cac.sigma = 0.5
parameter_list = original_gaussian_gray_cac.get_parameters()
dataset = original_gaussian_gray_cac.load_dataset(input_filename)
results_folder = output_folder + original_gaussian_gray_cac.CAC.__name__ + '/'
original_gaussian_gray_cac.test_model(dataset, parameter_list[0], results_folder, plot_evolution=True)
self.outside_seed_mean)
return grad_energy_1 + grad_energy_2
def _plotContourOnImage(self, contour_coord, cage_obj, alpha, grad_k, color=[0., 0., 255.]):
# hsi_image[:, :, 0] / (2 * 3.14) * 255.
utils.plotContourOnImage(contour_coord, self.image_obj.image,
points=cage_obj.cage,
color=color,
points2=cage_obj.cage - alpha * 10 * grad_k)
# self.image_obj.plot_hsi_image()
if __name__ == '__main__':
input_filename = 'BSDS300_input.txt'
output_folder = 'seg_im/'
color_cac = CACSegmenter(HueMeanSeedCAC)
parameter_list = color_cac.get_parameters()
new_parameters = {
'num_points': [12],
'ratio': [1.05],
'smallest_number': [np.exp(-200)],
}
color_cac.parameters = new_parameters
color_cac.sigma = 0.25
parameter_list = color_cac.get_parameters()
dataset = color_cac.load_dataset(input_filename)
results_folder = output_folder + color_cac.CAC.__name__+ '/'
color_cac.test_model(dataset, parameter_list[0], results_folder, plot_evolution=True)
omega2_coord, affine_omega_2_coord, self.outside_gmm, image,
image_gradient)
energy = omega_1 + omega_2
return energy
def _plotContourOnImage(self,
contour_coord,
cage_obj,
alpha,
grad_k,
color=[0., 0., 255.]):
utils.plotContourOnImage(contour_coord,
self.image_obj.image,
points=cage_obj.cage,
color=color,
points2=cage_obj.cage - alpha * 10 * grad_k)
if __name__ == '__main__':
input_filename = 'synthetic_input.txt'
output_folder = 'other/'
multivariate_gaussian_gray_cac = CACSegmenter(MultivariateGaussianCAC)
parameter_list = multivariate_gaussian_gray_cac.get_parameters()
dataset = multivariate_gaussian_gray_cac.load_dataset(input_filename)
results_folder = output_folder + multivariate_gaussian_gray_cac.CAC.__name__ + '/'
multivariate_gaussian_gray_cac.test_model(dataset,
parameter_list[0],
results_folder,
plot_evolution=True)
image = self.image_obj.gray_image
image_gradient = np.array(np.gradient(image))
# Calculate Energy:
omega_1 = mean_energy.mean_energy_grad_per_region(omega1_coord, affine_omega_1_coord, image, image_gradient)
omega_2 = mean_energy.mean_energy_grad_per_region(omega2_coord, affine_omega_2_coord, image, image_gradient)
energy = omega_1 + omega_2
return energy
def _plotContourOnImage(self, contour_coord, cage_obj, alpha, grad_k, color=[0., 0., 255.]):
utils.plotContourOnImage(contour_coord, self.image_obj.gray_image, points=cage_obj.cage, color=color,
points2=cage_obj.cage - alpha * 10 * grad_k)
if __name__ == '__main__':
input_filename = 'morphing_fruits_input.txt'
output_folder = 'segment_morphing_fruits/'
mean_gray_cac = CACSegmenter(MeanCAC)
new_parameters = {
'num_points': [10],
'ratio': [1.05],
}
mean_gray_cac.parameters = new_parameters
mean_gray_cac.sigma = 0.1
parameter_list = mean_gray_cac.get_parameters()
save_resulting_mask = '../../morphed_mask/'
dataset = mean_gray_cac.load_dataset(input_filename)
results_folder = output_folder + mean_gray_cac.CAC.__name__ + '/'
mean_gray_cac.test_model(dataset, parameter_list[0], results_folder, plot_evolution=True)
omega_2_coord, affine_omega_2_coord, self.image_obj)
return grad_energy_1 + grad_energy_2
def _plotContourOnImage(self,
contour_coord,
cage_obj,
alpha,
grad_k,
color=[0., 0., 255.]):
utils.plotContourOnImage(contour_coord,
self.image_obj.hsi_image[:, :, 0] /
(2 * 3.14) * 255.,
points=cage_obj.cage,
color=color,
points2=cage_obj.cage - alpha * 10 * grad_k)
# self.image_obj.plot_hsi_image()
if __name__ == '__main__':
input_filename = 'AlpertGBB07_input.txt'
output_folder = 'segment_results_alpert_3/'
color_cac = CACSegmenter(HueMeanCAC)
parameter_list = color_cac.get_parameters()
parameter_list = color_cac.get_parameters()
dataset = color_cac.load_dataset(input_filename)
results_folder = output_folder + color_cac.CAC.__name__ + '/'
color_cac.test_model(dataset,
parameter_list[0],
results_folder,
plot_evolution=True)
grad_energy = grad_energy_1 + grad_energy_2
total_grad_energy.append(w * grad_energy)
return sum(total_grad_energy)
def _plotContourOnImage(self,
contour_coord,
cage_obj,
alpha,
grad_k,
color=[0., 0., 255.]):
utils.plotContourOnImage(contour_coord,
self.image_obj.image,
points=cage_obj.cage,
color=color,
points2=cage_obj.cage - alpha * 10 * grad_k)
if __name__ == '__main__':
input_filename = 'AlpertGBB07_input.txt'
output_folder = 'segment_results_alpert_4/'
rgb_cac = CACSegmenter(MeanMultiCAC,
type=['N', 'N', 'N'],
weight=[0, 0, 1])
parameter_list = rgb_cac.get_parameters()
dataset = rgb_cac.load_dataset(input_filename)
results_folder = output_folder + rgb_cac.CAC.__name__ + '/'
rgb_cac.test_model(dataset,
parameter_list[0],
results_folder,
plot_evolution=True)
energy = (omega_1 + omega_2) / 2.
return energy
def energy_gradient(self, omega1_coord, omega2_coord, affine_omega_1_coord, affine_omega_2_coord):
# Calculate Image gradient
image = self.image_obj.image
image_gradient = np.array([np.array(np.gradient(image[:, :, slice])) for slice in xrange(image.shape[2])])
image_gradient = np.transpose(image_gradient, (1,2,3,0))
# Calculate Energy:
omega_1 = g_energies.grad_gauss_energy_per_region(omega1_coord, affine_omega_1_coord, self.inside_gmm, image,
image_gradient)
omega_2 = g_energies.grad_gauss_energy_per_region(omega2_coord, affine_omega_2_coord, self.outside_gmm, image,
image_gradient)
energy = omega_1 + omega_2
return energy
def _plotContourOnImage(self, contour_coord, cage_obj, alpha, grad_k, color=[0., 0., 255.]):
utils.plotContourOnImage(contour_coord, self.image_obj.image, points=cage_obj.cage, color=color,
points2=cage_obj.cage - alpha * 10 * grad_k)
if __name__ == '__main__':
input_filename = 'synthetic_input.txt'
output_folder = 'other/'
multivariate_gaussian_gray_cac = CACSegmenter(MultivariateGaussianCAC)
parameter_list = multivariate_gaussian_gray_cac.get_parameters()
dataset = multivariate_gaussian_gray_cac.load_dataset(input_filename)
results_folder = output_folder + multivariate_gaussian_gray_cac.CAC.__name__ +'/'
multivariate_gaussian_gray_cac.test_model(dataset, parameter_list[0], results_folder, plot_evolution=True)
energy = (omega_1 + omega_2) / float(2)
total_energy.append(w * energy)
return sum(total_energy)
def energy_gradient(self, omega_1_coord, omega_2_coord, affine_omega_1_coord, affine_omega_2_coord):
total_grad_energy = []
for slice, (t, w) in enumerate(zip(self.type, self.weight)):
grad_energy_1 = multi_mean_energy.generic_grad_mean_energy_per_region(omega_1_coord, affine_omega_1_coord,
self.image_obj, t,
slice)
grad_energy_2 = multi_mean_energy.generic_grad_mean_energy_per_region(omega_2_coord, affine_omega_2_coord,
self.image_obj, t,
slice)
grad_energy = grad_energy_1 + grad_energy_2
total_grad_energy.append(w * grad_energy)
return sum(total_grad_energy)
def _plotContourOnImage(self, contour_coord, cage_obj, alpha, grad_k, color=[0., 0., 255.]):
utils.plotContourOnImage(contour_coord, self.image_obj.image, points=cage_obj.cage, color=color,
points2=cage_obj.cage - alpha * 10 * grad_k)
if __name__ == '__main__':
input_filename = 'AlpertGBB07_input.txt'
output_folder = 'segment_results_alpert_4/'
rgb_cac = CACSegmenter(MeanMultiCAC, type=['N', 'N', 'N'], weight=[0, 0, 1])
parameter_list = rgb_cac.get_parameters()
dataset = rgb_cac.load_dataset(input_filename)
results_folder = output_folder + rgb_cac.CAC.__name__ + '/'
rgb_cac.test_model(dataset, parameter_list[0], results_folder, plot_evolution=True)