def folds_annot():
train_data = get_10lamb_old(5)
img_x, _, _, _ = get_training_data(train_data)
img_clean = img_x[..., :3]
lst_get = [get_borders1, get_borders2, get_borders3, get_borders4, get_borders5, get_borders6]
for i_fold in range(6):
img_annot = imread(f'/home/lameeus/data/ghent_altar/input/hierachy/10_lamb/annotations/kfold/annot_{i_fold+1}.png')
y1 = annotations2y(img_annot, thresh=.8)[..., 1]
a = semi_transparant(img_clean, y1.astype(bool))
w0, w1, h0, h1 = lst_get[i_fold]()
clean_annot_crop = a[h0:h1, w0:w1, :]
img_clean_crop = img_clean[h0:h1, w0:w1, :]
if 0: concurrent([img_clean_crop, clean_annot_crop])
folder_save = '/scratch/lameeus/data/ghent_altar/input/hierarchy/10lamb/ifolds'
imsave(os.path.join(folder_save, f'clean_crop_ifold{i_fold}.png'), img_clean_crop)
imsave(os.path.join(folder_save, f'clean_annot_crop_ifold{i_fold}.png'), clean_annot_crop)
pass
def predict_average(self):
y_lst = []
for i_fold in range(6):
for k in range(18, 19 + 1):
for epoch in range(99, 100 + 1):
y = self.predict(i_fold=i_fold, k=k, epoch=epoch)
y_lst.append(y)
y_avg = np.mean(y_lst, axis=0)
y_avg_bin = self.get_bin(y_avg)
img_clean = self.img_x[..., :3]
y_avg_bin_overlay = semi_transparant(img_clean, y_avg_bin)
concurrent([y_avg[..., 1], y_avg_bin, y_avg_bin_overlay, img_clean])
if 0:
path_save = f'/scratch/lameeus/data/ghent_altar/output/hierarchy/10_lamb/' \
f'paintloss_tiunet_enc{self.fixed_enc}.png'
imsave(path_save, y_avg_bin)
if 1:
path_save = f'/scratch/lameeus/data/ghent_altar/output/hierarchy/10_lamb/' \
f'paintloss_overlay_tiunet_enc{self.fixed_enc}.png'
imsave(path_save, y_avg_bin_overlay)
return 1
def plot(self):
imgs = self.get()
keys = imgs.keys()
img_list = [imgs[key] for key in keys]
concurrent(img_list, titles=keys)
def predict_compare_regular(self):
img_y_all = self.k_fold_train_data.get_train_data_all().get_y_train()
lst_get = [
get_borders1, get_borders2, get_borders3, get_borders4,
get_borders5, get_borders6
]
img_clean = self.img_x[..., :3]
for i_fold in range(6):
y_lst = []
# Average prediction
print(f'i_fold = {i_fold}')
for k in [17, 18, 19]:
print(f'k = {k}')
for epoch in [36, 37, 38, 39, 40]:
print(f'epoch = {epoch}')
y = self.predict_regular(i_fold=i_fold, k=k, epoch=epoch)
y_lst.append(y)
y_avg = np.mean(y_lst, axis=0)
y_avg_bin = self.get_bin(y_avg)
# Performance
img_y_te = self.k_fold_train_data.k_split_i(i_fold).get_y_test()
thresh = optimal_test_thresh_equal_distribution(img_y_all, y_avg)
perf = foo_performance(img_y_te, y_avg, thresh)
# CROP
w0, w1, h0, h1 = lst_get[i_fold]()
y_avg_bin_crop = y_avg_bin[h0:h1, w0:w1]
clean_crop = img_clean[h0:h1, w0:w1, :]
y_avg_bin_transparent_crop = semi_transparant(
clean_crop, y_avg_bin_crop)
if 0:
concurrent(
[clean_crop, y_avg_bin_crop, y_avg_bin_transparent_crop])
folder_save = '/scratch/lameeus/data/ghent_altar/output/hierarchy/10_lamb/ifolds_regular_tiunet'
filename = f'_tiunet_ifold{i_fold}_jacc{perf["jaccard"]:.3f}.png'
# Save y_bin
imsave(os.path.join(folder_save, 'binpred' + filename),
y_avg_bin_crop,
b_check_duplicate=False)
# Save overlay
imsave(os.path.join(folder_save, 'overlay' + filename),
y_avg_bin_transparent_crop,
b_check_duplicate=False)
def predict_regular(self, i_fold=None, k=None, epoch=None):
n = self.get_n_model_regular(i_fold=i_fold, k=k, epoch=epoch)
y = n.predict(self.img_x)
if 0:
concurrent([y[..., 1], self.img_x[..., :3]])
return y
def local_thresholding(x_img, ext:int=200):
gray = np.mean(x_img, axis=2)
x_threshold = filters.threshold_local(gray, block_size=ext*2+1)
# Debugging:
if 1:
from plotting import concurrent
concurrent([gray, x_threshold])
return np.greater_equal(gray, x_threshold)
def __init__(self):
# data
self.data()
# Load model(s)
model_name = 'unet' # ['simple', 'ti-unet', 'unet']:
folder = f'C:/Users/admin/Data/ghent_altar/net_weight/{model_name}_d1_k9_n80'
epoch = 1
path = f'C:/Users/admin/Data/ghent_altar/net_weight/{model_name}_d1_k9_n80/w_{epoch}.h5'
from scripts.scripts_performance.main_performance import load_model_quick
model = load_model_quick(path)
neural_net = NeuralNet(model, w_ext=10, norm_x=True)
model.summary()
for epoch in range(1, 10 + 1):
print('epoch', epoch)
neural_net.load(folder, epoch)
# Predict
y_pred = neural_net.predict(self.img_x)
if 0:
plt.imshow(y_pred[..., 0])
plt.show()
for val_name in self.val:
print(val_name)
y_true_val = self.val[val_name]
data_i = _eval_func_single(y_true_val, y_pred)
print(data_i)
# TODO best performing (ti-unet: 4)
neural_net.load(folder, 4)
y_pred = neural_net.predict(self.img_x)
from performance.testing import get_y_pred_thresh
y_pred_thresh = get_y_pred_thresh(y_pred, data_i['thresh'])
concurrent([
self.img_x[..., :3], self.img_y[..., 0], y_pred[..., 0],
y_pred_thresh[..., 0]
])
y_pred
def ae_results(self, n_ae):
"""
AE Reconstruction
"""
img_x_ae = n_ae.predict(self.img_x)
y_denorm = _undo_norm_input(img_x_ae)
y_clean = y_denorm[..., :3]
y_rgb = y_denorm[..., 3:6]
y_ir = y_denorm[..., 6]
y_irr = y_denorm[..., 7]
y_xray = y_denorm[..., 8]
concurrent([y_clean, y_rgb, y_ir, y_irr, y_xray],
['clean', 'rgb', 'ir', 'irr', 'xray'])
def main_eval(
self,
train_data,
b_plot=False,
):
x = train_data.get_x_test()
y_pred = self.neural_net.predict(x)
if b_plot:
concurrent([x[..., :3], y_pred[..., 1]], ['input', 'prediction'])
val_datas = [{'y': train_data.get_y_test()}] + self.val_datas
return _eval_func(y_pred, val_datas, b_plot=b_plot)
def continues_learning():
folder = '/home/lameeus/data/ghent_altar/input/hierarchy/13_small'
im_clean = imread(os.path.join(folder, 'clean.png'))[..., :3]
im_annot0 = imread(os.path.join(folder, 'annot.tif'))
im_annot1 = imread(os.path.join(folder, 'clean_annot_practical.png'))
y_true = annotations2y(im_annot0)
y_true_extra = annotations2y(im_annot1, thresh=.9)
folder = '/home/lameeus/data/ghent_altar/output/hierarchy/13_small/practical_annotations'
y_pred0 = imread(os.path.join(folder, 'pred_transfer_kfoldenc2_ifold0_avg.png'))
folder = '/home/lameeus/data/ghent_altar/output/hierarchy/13_small'
y_pred1 = imread(os.path.join(folder, 'pred_transfer_kfoldenc2_ifold0_avg_epoch50_J0427.png'))
from performance.testing import optimal_test_thresh_equal_distribution
from scripts.scripts_performance.main_performance import foo_performance
from figures_paper.overlay import semi_transparant
def get_bin(y_pred):
assert len(y_pred.shape) == 2
y_pred01 = np.stack([1-y_pred, y_pred], axis=-1)
thresh = optimal_test_thresh_equal_distribution(y_true, y_pred01)
print(foo_performance(y_true, y_pred01, thresh))
y_pred_bin = y_pred >= thresh
return y_pred_bin
y_pred0_bin = get_bin(y_pred0)
y_pred1_bin = get_bin(y_pred1)
y_pred0_bin_fancy = semi_transparant(im_clean, y_pred0_bin)
y_pred1_bin_fancy = semi_transparant(im_clean, y_pred1_bin)
concurrent([im_clean, y_true[..., 0], y_true_extra[..., 0], y_pred0, y_pred1, y_pred0_bin, y_pred1_bin,
y_pred0_bin_fancy,
y_pred1_bin_fancy])
folder_save = '/home/lameeus/data/ghent_altar/output/hierarchy/13_small/fancy'
imsave(os.path.join(folder_save, 'overalytrain10.png'), y_pred0_bin_fancy)
imsave(os.path.join(folder_save, 'overalytrain10train13.png'), y_pred1_bin_fancy)
return
def pred_epochs():
img_x, img_y_val = data_lamb()
d = 2
k = 10
model_name = 'ti-unet'
train_data = '1319_10nat'
w_ext = 10 if d == 1 else 26
y_pred_lst = []
n = []
for epoch in range(10, 101, 10):
print(epoch)
epoch_start = 50
epoch_corr = epoch + epoch_start if train_data[:5] == '1319_' else epoch
path = f'C:/Users/admin/Data/ghent_altar/net_weight/{train_data}/{model_name}_d{d}_k{k}/w_{epoch_corr}.h5'
try:
model = load_model_quick(path)
except Exception as e:
print(e)
continue
neural_net = NeuralNet(model, w_ext=w_ext, norm_x=True)
y_pred = neural_net.predict(img_x)
if 0:
data_i = _eval_func_single(img_y_val, y_pred, metric='kappa')
print(data_i)
data_i = _eval_func_single(img_y_val, y_pred, metric='jaccard')
print(data_i)
y_pred_lst.append(y_pred)
n.append(epoch)
concurrent([a[..., 1] for a in y_pred_lst], n)
plt.show()
return 1
def foo(n_segm, b=0):
y_pred = n_segm.predict(self.img_x)
thresh_single = optimal_test_thresh_equal_distribution(
self.img_y_te, y_pred)
# data_single_i = {'k': self.k,
# 'i_fold': i_fold,
# 'epoch': epoch}
print(foo_performance(self.img_y_te, y_pred, thresh_single))
img_clean = self.img_x[..., :3]
concurrent([
img_clean, y_pred[..., 1], y_pred[..., 1] >= thresh_single,
semi_transparant(img_clean, y_pred[..., 1] >= thresh_single)
])
if b:
from data.datatools import imsave
folder = '/home/lameeus/data/ghent_altar/output/hierarchy/'
info_epoch = f'_epoch{n_segm.epoch}' if n_segm.epoch > 0 else ''
filename = folder + f'13_small/pred_transfer_kfoldenc{self.fixed_enc}_ifold{self.i_fold}_avg{info_epoch}.png'
imsave(filename, y_pred[..., 1])
def main():
### Settings
mod=5
panel_nr = 19
i_start ,i_end = 1, epochs_tot
# i_start ,i_end = 1, 2
k_lst = np.arange(1, 21)
# k_lst = [1, 2]
verbose=0
b_plot = False
###
if panel_nr == 13:
train_data = get_13botleftshuang(mod=mod)
folder_weights = '/scratch/lameeus/data/ghent_altar/net_weight/lamb_segmentation'
elif panel_nr == 19:
train_data = get_19SE_shuang(mod=mod)
folder_weights = '/scratch/lameeus/data/ghent_altar/net_weight/19_hand_SE'
else:
raise ValueError(panel_nr)
x, y_tr, _, y_te = get_training_data(train_data)
(y_tr, y_te) = map(batch2img, (y_tr, y_te))
assert i_end >= i_start
if b_plot:
# plotting
pred_lst = []
info_lst = []
lst_data = []
lst_data_avg_pred = []
for k in k_lst:
model = None
pred_lst = []
for epoch in np.arange(i_start, i_end + 1)[::-1]:
info = f'settings: k {k}; epoch {epoch}'
print('\n\t'+info)
filepath_model = os.path.join(folder_weights, f'ti_unet_k{k}_imbalanced/w_{epoch}.h5')
if epoch == i_end:
model = load_model(filepath_model, custom_objects={'loss': loss,
'accuracy_with0': accuracy_with0,
'jaccard_with0': jaccard_with0,
'kappa_loss': kappa_loss
})
else:
model.load_weights(filepath_model)
n = NeuralNet(model, w_ext=10)
y_pred = n.predict(x)
o = y_pred[..., 1]
pred_lst.append(o)
def print_conf(y_true, y_pred):
y_true = batch2img(y_true)
y_pred = batch2img(y_pred)
b_annot = np.sum(y_true, axis=-1).astype(bool)
y_true_annot = y_true[b_annot, :].argmax(axis=-1)
y_pred_annot = y_pred[b_annot, :].argmax(axis=-1)
"""
T0; predicted 1, but is 0
predicted 0, but is 1; T1
"""
conf_mat = confusion_matrix(y_true_annot, y_pred_annot)
print(conf_mat)
if 1: # Single prediction
if verbose == 1:
print_conf(y_tr, y_pred)
print_conf(y_te, y_pred)
if b_plot:
pred_lst.append(o)
info_lst.append(info)
test_thresh = test_thresh_incremental(y_pred, y_tr, y_te, n=5, verbose=0)
pred_thresh = np.greater_equal(o, test_thresh)
pred_thresh_bin = np.stack([1-pred_thresh, pred_thresh], axis=-1)
y_te_flat, y_pred_flat = filter_non_zero(y_te, pred_thresh_bin)
y_te_argmax = np.argmax(y_te_flat, axis=-1)
y_pred_argmax = np.argmax(y_pred_flat, axis=-1)
acc, jacc, kappa = _get_scores(y_te_argmax, y_pred_argmax)
if verbose == 1:
print_conf(y_tr, pred_thresh_bin)
print_conf(y_te, pred_thresh_bin)
if 0: concurrent([pred_thresh])
data_i = {'k':k,
'epoch':epoch,
'test_thresh':test_thresh,
'kappa':kappa,
'accuracy':acc,
'jaccard':jacc
}
lst_data.append(data_i)
if 1: # avg prediction
pred_i_average = np.mean(pred_lst, axis=0)
# optimizing threshold prediction
test_thresh = test_thresh_incremental(np.stack([1 - pred_i_average, pred_i_average], axis=-1), y_tr, y_te, n=5,
verbose=0)
pred_thresh = np.greater_equal(pred_i_average, test_thresh)
pred_thresh_bin = np.stack([1 - pred_thresh, pred_thresh], axis=-1)
y_te_flat, y_pred_flat = filter_non_zero(y_te, pred_thresh_bin)
y_te_argmax = np.argmax(y_te_flat, axis=-1)
y_pred_argmax = np.argmax(y_pred_flat, axis=-1)
acc, jacc, kappa = _get_scores(y_te_argmax, y_pred_argmax)
data_i = {'k': k,
'epoch_start': epoch,
'test_thresh': test_thresh,
'kappa': kappa,
'accuracy': acc,
'jaccard': jacc
}
lst_data_avg_pred.append(data_i)
b = True
if b:
df = pd.DataFrame(lst_data)
filename_save = f'tiunet_1pool_shaoguang{panel_nr}_imbalanced'
filename_path = f'/scratch/lameeus/data/ghent_altar/dataframes/{filename_save}.csv'
df.to_csv(filename_path, sep=';')
df = pd.DataFrame(lst_data_avg_pred)
filename_save = f'tiunet_1pool_shaoguang{panel_nr}_imbalanced_averaging'
df.to_csv(f'/scratch/lameeus/data/ghent_altar/dataframes/{filename_save}.csv', sep=';')
if b_plot:
concurrent(pred_lst, info_lst)
plt.show()
return
except Exception as e:
print(e)
continue
y_pred_lst.append(y_pred_i[..., 1])
y_pred_avg = np.mean(y_pred_lst, axis=0)
return y_pred_avg
if 1:
# Average out prediction
y_pred_avg = average_out_pred()
if 0:
concurrent([y_pred_avg])
y_pred_avg2 = np.stack([1 - y_pred_avg, y_pred_avg], axis=-1)
data_i = _eval_func_single(get_crop(img_y), get_crop(y_pred_avg2))
print(data_i)
from performance.testing import get_y_pred_thresh
thresh = df.iloc[i_max]['thresh']
y_thresh = get_y_pred_thresh(y_pred, thresh=thresh)
concurrent([img_x[..., :3], y_pred[..., 0], y_thresh[..., 0]])
im_pred = get_crop(y_thresh)
def predict_compare(self):
img_y_all = self.k_fold_train_data.get_train_data_all().get_y_train()
lst_get = [
get_borders1, get_borders2, get_borders3, get_borders4,
get_borders5, get_borders6
]
img_clean = self.img_x[..., :3]
for i_fold in range(6):
for i_fixed_enc in range(3):
self.fixed_enc = i_fixed_enc
y_lst = []
# Average prediction
for k in [17, 18, 19]:
for epoch in [96, 97, 98, 99, 100]:
y = self.predict(i_fold=i_fold, k=k, epoch=epoch)
y_lst.append(y)
y_avg = np.mean(y_lst, axis=0)
y_avg_bin = self.get_bin(y_avg)
# Performance
img_y_te = self.k_fold_train_data.k_split_i(
i_fold).get_y_test()
thresh = optimal_test_thresh_equal_distribution(
img_y_all, y_avg)
perf = foo_performance(img_y_te, y_avg, thresh)
# CROP
w0, w1, h0, h1 = lst_get[i_fold]()
y_avg_bin_crop = y_avg_bin[h0:h1, w0:w1]
clean_crop = img_clean[h0:h1, w0:w1, :]
y_avg_bin_transparent_crop = semi_transparant(
clean_crop, y_avg_bin_crop)
if 0:
concurrent([
clean_crop, y_avg_bin_crop, y_avg_bin_transparent_crop
])
# Save
if self.fixed_enc == 0: info_enc = 'Train'
elif self.fixed_enc == 1: info_enc = 'Fixed'
elif self.fixed_enc == 2: info_enc = 'FixedTrain'
folder_save = '/scratch/lameeus/data/ghent_altar/output/hierarchy/10_lamb/ifolds'
filename = f'_enc{info_enc}_ifold{i_fold}_jacc{perf["jaccard"]:.3f}.png'
# Save y_bin
imsave(os.path.join(folder_save, 'binpred' + filename),
y_avg_bin_crop,
b_check_duplicate=False)
# Save overlay
imsave(os.path.join(folder_save, 'overlay' + filename),
y_avg_bin_transparent_crop,
b_check_duplicate=False)
def transfer_learning(
epoch=25, # Could check a few
b_plot=False):
d = 2 # 1, 2
img_x, img_y_val = data_lamb()
k = 10
model_name = 'ti-unet'
w_ext = 10 if d == 1 else 26
# train_data:
y_pred_lst = []
n = ['clean']
# train_data_lst = ['1319_10', '10', '1319', '1319_101319']
train_data_lst = ['10nat', '1319_10nat', '1319_10nat1319', '1319']
data_i_lst = {}
for train_data in train_data_lst:
print(train_data)
epoch_start = 50
epoch_corr = epoch + epoch_start if train_data[:5] == '1319_' else epoch
if train_data == '1319':
epoch_corr = 50
path = f'C:/Users/admin/Data/ghent_altar/net_weight/{train_data}/{model_name}_d{d}_k{k}/w_{epoch_corr}.h5'
try:
model = load_model_quick(path)
except Exception as e:
print(e)
continue
neural_net = NeuralNet(model, w_ext=w_ext, norm_x=True)
y_pred = neural_net.predict(img_x)
# baseline
data_i = _eval_func_single(img_y_val, y_pred, metric='kappa')
print(data_i)
if 0:
"""
Checking which
baseline ~ .22
i = 0: .268, Remove huge improvement ( a lot of "green" background annotated as paint loss)
i = 1: .228 Keep!
i = 2: .179 keep! Drop (keep!!
i = 3: .159 keep! Even more important
i = 4: .252 Remove (huge problem right top)
i = 5: .233 Keep, quit relevant
"""
from datasets.default_trainingsets import get_10lamb_6patches
kFoldTrainData = get_10lamb_6patches(5)
_eval_func_single(
kFoldTrainData.k_split_i(0).get_y_train(),
y_pred,
metric='kappa') # Check what is influence without!
data_i_lst[train_data] = data_i
data_i = _eval_func_single(img_y_val, y_pred, metric='jaccard')
print(data_i)
y_pred_lst.append(y_pred)
n.append(train_data)
# plt.imshow(neural_net.predict(img_x[::2,::2,:])[..., 1])
if b_plot:
concurrent([img_x[..., :3]] + [a[..., 1] for a in y_pred_lst], n)
if 0:
from figures_paper.overlay import semi_transparant
from data.datatools import imread, imsave
t = [data_i_lst[n_i]['thresh'] for n_i in train_data_lst]
p = []
for i, train_data in enumerate(train_data_lst):
b = np.greater_equal(y_pred_lst[i][..., 1], t[i])
k = semi_transparant(img_x[..., :3], b, 0)
p.append(k)
imsave(
os.path.join(
"C:/Users/admin/OneDrive - ugentbe/data/images_paper",
train_data + '.png'), k)
concurrent(p)
return data_i_lst
if 0:
a = get_19hand()
b = False
if b:
a.plot()
### Training/Validation data
img_y = a.get('annot')
y = annotations2y(img_y)
y_annot = y2bool_annot(y)
b = False
if b:
y_annot_tr, y_annot_te = panel19withoutRightBot(y_annot)
concurrent([a.get('clean'), y_annot, y_annot_tr, y_annot_te],
['clean', 'annotation', 'a annot', 'test annot'])
if 0:
train_data = get_train19_topleft(mod=mod)
elif dataset_name == '19_hand_SE':
train_data = get_19SE_shuang(mod=mod)
else:
train_data = get_13botleftshuang(mod=mod)
# TODO normalise inputs This seems to be super important...
# train_data.x = (1/255. * train_data.x).astype(np.float16)
# train_data.x = (255. * train_data.x).astype(np.float16)
x, y_tr, x_te, y_te = get_training_data(train_data)
# To get w_ext
def main():
b_encoder_fixed = False
info_enc_fixed = '_enc_fixed'
folder_weights = '/scratch/lameeus/data/ghent_altar/net_weight/10lamb_kfold_pretrained'
folder_save = '/home/lameeus/data/ghent_altar/dataframes'
filename_single = f'pretrained_unet_10lamb_kfold_single'
filename_avg_pred = f'pretrained_unet_10lamb_kfold_avgpred'
folder_weights += info_enc_fixed if b_encoder_fixed else ''
filename_single += info_enc_fixed if b_encoder_fixed else ''
filename_avg_pred += info_enc_fixed if b_encoder_fixed else ''
fold_range = range(6)
# fold_range = [0, 1]
k = 10
epoch_range = range(1, 40 + 1)
w_ext_in = 28
k_fold_train_data = get_10lamb_6patches(5) # 5 is the number of modalities
train_data_all = k_fold_train_data.get_train_data_all()
img_x = train_data_all.get_x_train()
img_x = rescale0to1(img_x)
img_clean = img_x[..., :3]
img_y_all = train_data_all.get_y_train()
b_plot = False
for i_fold in fold_range:
print(i_fold)
img_y_te = k_fold_train_data.k_split_i(i_fold).get_y_test()
# Init for range epochs
lst_data_single = []
lst_data_avg_pred = []
list_y_pred = []
model = None
for epoch in np.sort(epoch_range)[::-1]:
filepath_model = os.path.join(
folder_weights, f'unet_enc_k{k}_ifold{i_fold}/w_{epoch}.h5')
model = load_model_quick(filepath_model, model=model)
n = NeuralNet(model, w_ext=w_ext_in)
y_pred = n.predict(img_x)
"""
Average out predictions
"""
list_y_pred.append(y_pred)
y_avg_pred = np.mean(list_y_pred, axis=0)
thresh_single = optimal_test_thresh_equal_distribution(
img_y_all, y_pred)
thresh_avg_pred = optimal_test_thresh_equal_distribution(
img_y_all, y_avg_pred)
y_pred_bin = np.greater_equal(y_pred[..., 1], thresh_single)
dict_perf = foo_performance(img_y_te, y_pred, thresh_single)
print(dict_perf)
if b_plot:
concurrent([
y_pred_bin, img_clean,
semi_transparant(img_clean, y_pred_bin),
semi_transparant(img_clean, img_y_te[..., 1].astype(bool))
])
data_single_i = {'k': k, 'i_fold': i_fold, 'epoch': epoch}
data_avg_pred_i = {
'k': k,
'i_fold': i_fold,
'epoch_start': epoch,
'epoch_end': max(epoch_range)
}
data_single_i.update(dict_perf)
data_avg_pred_i.update(
foo_performance(img_y_te, y_avg_pred, thresh_avg_pred))
lst_data_single.append(data_single_i)
lst_data_avg_pred.append(data_avg_pred_i)
df_single = pd.DataFrame(lst_data_single)
df_avg_pred = pd.DataFrame(lst_data_avg_pred)
path_single = os.path.join(folder_save, filename_single + '.csv')
path_avg_pred = os.path.join(folder_save, filename_avg_pred + '.csv')
pandas_save(path_single, df_single, append=True)
pandas_save(path_avg_pred, df_avg_pred, append=True)
return
# Read paint loss detection
b_mask = detect_colour(
imread(
"C:/Users/admin/OneDrive - ugentbe/data/images_paper/1319_10nat_V3.png"
), "cyan")
if 0:
plt.imshow(b_mask)
# Read inpaint
im_inpaint = imread(os.path.join(f, "inpainting_comb.jpg"))
im_inpainted_new = inpaint_replacer(im_orig, b_mask, im_inpaint)
plt.imshow(im_inpainted_new)
concurrent([im_orig, im_inpainted_new, im_inpaint, b_mask],
['orig', 'new', 'old', 'mask'])
# Save
imsave(os.path.join(f, "inpaint_v2.png"), im_inpainted_new)
# Crop?
im_treated = imread(os.path.join(f, "restored_SR.jpg"))
# Trying to rescale to similar colour pallet
im_treated_recolour = im_treated.astype(float)
im_treated_recolour = (im_treated_recolour - im_treated.mean(
(0, 1))) * im_inpainted_new.std((0, 1)) / im_treated.std(
(0, 1)) + im_inpainted_new.mean((0, 1))
im_treated_recolour = np.clip(im_treated_recolour, 0, 255)
im_treated_recolour = im_treated_recolour.astype(np.uint8)
def main():
"""
:return:
"""
### Settings
mod = 5
w_patch = 16 * 2
"""
Data (all important modalities)
"""
# folder_windows = r'C:\Users\Laurens_laptop_w\OneDrive - UGent\data\10lamb'
train_data = get_10lamb_old(mod)
img_x, img_y_tr, _, _ = get_training_data(train_data)
# Normalise the input!
img_x = rescale0to1(img_x)
"""
Train segmentation
1) reuse everything
2) fix encoder
"""
if 1:
if 1:
b_encoder_fixed = False
info_enc_fixed = '_enc_fixed' if b_encoder_fixed else ''
get_info = lambda: f'10lamb_kfold_pretrained{info_enc_fixed}/unet_enc_k{k}_ifold{i_fold}'
n_epochs = 40
k = 10
if k == 10:
epoch_w = 100
else:
raise NotImplementedError()
### Settings you don't have to change:
w_patch = 50
w_ext_in = 28
b_double = False
padding = 'valid'
# TODO flag for converting encoder to dilated conv
def get_unet_pretrained_encoder():
model_encoder = get_model_encoder()
encoder_inputs = model_encoder.input
decoder_outputs = decoder(model_encoder, f_out=2)
model_pretrained_unet = Model(encoder_inputs, decoder_outputs)
from methods.examples import compile_segm
compile_segm(model_pretrained_unet, lr=1e-4)
model_pretrained_unet.summary()
return model_pretrained_unet
"""
Train
"""
k_fold_train_data = get_10lamb_6patches(5)
for i_fold in range(6):
"""
Get a new network (not trained yet for segmentation)
"""
model_pretrained_unet = get_unet_pretrained_encoder()
n_pretrained_unet = NeuralNet(model_pretrained_unet)
"""
The data
"""
train_data_i = k_fold_train_data.k_split_i(i_fold)
info = get_info()
img_y_tr = train_data_i.get_y_train()
img_y_te = train_data_i.get_y_test()
flow_tr = get_flow(
img_x,
img_y_tr,
w_patch=w_patch, # Comes from 10
w_ext_in=w_ext_in)
flow_te = get_flow(
img_x,
img_y_te,
w_patch=w_patch, # Comes from 10
w_ext_in=w_ext_in)
n_pretrained_unet.train(flow_tr,
flow_te,
epochs=n_epochs,
verbose=1,
info=info)
"""
Prediction
"""
n_pretrained_unet.w_ext = w_ext_in
y_pred = n_pretrained_unet.predict(img_x)
concurrent([y_pred[..., 1]])
"""
Classification
"""
if 1:
im_clean = img_x[..., :3]
k = 8
i_fold = 3
epoch_last = 40
from methods.examples import kappa_loss, weighted_categorical_crossentropy
from performance.metrics import accuracy_with0, jaccard_with0
loss = weighted_categorical_crossentropy((1, 1))
list_y_pred = []
### K fold validation
k_fold_train_data = get_10lamb_6patches(5)
train_data_i = k_fold_train_data.k_split_i(i_fold)
img_y_tr = train_data_i.get_y_train()
img_y_te = train_data_i.get_y_test()
for epoch in np.arange(31, epoch_last + 1):
filepath_model = f'/scratch/lameeus/data/ghent_altar/net_weight/10lamb_kfold/ti_unet_k{k}_kfold{i_fold}/w_{epoch}.h5'
model = load_model(filepath_model,
custom_objects={
'loss': loss,
'accuracy_with0': accuracy_with0,
'jaccard_with0': jaccard_with0,
'kappa_loss': kappa_loss
})
n = NeuralNet(model, w_ext=10)
y_pred = n.predict(img_x)
list_y_pred.append(y_pred)
y_pred_mean = np.mean(list_y_pred, axis=0)
q1 = y_pred_mean[..., 1]
concurrent([q1, q1.round(), im_clean])
"""
Optimal threshold (making conf matrix symmetric, not based on maximising kappa)
"""
y_gt = np.any([img_y_tr, img_y_te], axis=0)
from performance.testing import _get_scores, filter_non_zero
def foo_performance(y_true, y_pred, thresh):
# is basically argmax
y_pred_thresh_arg = np.greater_equal(y_pred[..., 1], thresh)
y_true_flat, y_pred_thresh_arg_flat = filter_non_zero(
y_true, y_pred_thresh_arg)
y_te_argmax = np.argmax(y_true_flat, axis=-1)
# Kappa
return _get_scores(y_te_argmax, y_pred_thresh_arg_flat)[-1]
"""
1. BEST? PERFORMANCE based on test set
"""
print('1. Test distribution optimization')
thresh = optimal_test_thresh_equal_distribution(img_y_te, y_pred_mean)
q1_thresh = np.greater_equal(q1, thresh)
concurrent([q1, q1_thresh, im_clean])
print(f'thresh: {thresh}')
# Test, train, both
print('Kappa performance:')
print('\ttrain:', foo_performance(img_y_tr, y_pred_mean, thresh))
print('\ttestset:', foo_performance(img_y_te, y_pred_mean, thresh))
print('\tboth:', foo_performance(y_gt, y_pred_mean, thresh))
print('\nIncremental optimization on test set')
test_thresh2 = test_thresh_incremental(y_pred_mean,
img_y_tr,
img_y_te,
n=5,
verbose=0)
print('Kappa performance:')
print('\ttrain:', foo_performance(img_y_tr, y_pred_mean, test_thresh2))
print('\ttestset:', foo_performance(img_y_te, y_pred_mean,
test_thresh2))
print('\tboth:', foo_performance(y_gt, y_pred_mean, test_thresh2))
"""
2. based on train
"""
print('\n2. Training distribution optimization')
thresh = optimal_test_thresh_equal_distribution(img_y_tr, y_pred_mean)
q1_thresh = np.greater_equal(q1, thresh)
concurrent([q1, q1_thresh, im_clean])
print(f'thresh: {thresh}')
# Test, train, both
print('Kappa performance:')
print('\ttrain:', foo_performance(img_y_tr, y_pred_mean, thresh))
print('\ttestset:', foo_performance(img_y_te, y_pred_mean, thresh))
print('\tboth:', foo_performance(y_gt, y_pred_mean, thresh))
"""
3. CONSISTENT: based on train+set
"""
print('\n3. all GT distribution optimization')
thresh = optimal_test_thresh_equal_distribution(y_gt, y_pred_mean)
q1_thresh = np.greater_equal(q1, thresh)
concurrent([q1, q1_thresh, im_clean])
print(f'thresh: {thresh}')
# Test, train, both
print('Kappa performance:')
print('\ttrain:', foo_performance(img_y_tr, y_pred_mean, thresh))
print('\ttestset:', foo_performance(img_y_te, y_pred_mean, thresh))
print('\tboth:', foo_performance(y_gt, y_pred_mean, thresh))
if 0:
"""
4. DUMB/Not needed: Based on prediction of whole panel
"""
thresh = optimal_test_thresh_equal_distribution(y_gt,
y_pred_mean,
mask_true=False)
q1_thresh = np.greater_equal(q1, thresh)
concurrent([q1, q1_thresh, im_clean])
print('Done')