Skip to content

Sofia #205

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

Sofia #205

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a3da524
add print statements to face detection folder
SofiaVega Nov 18, 2021
56f4b88
change train_domain_A
SofiaVega Nov 24, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 17 additions & 5 deletions Face_Detection/align_warp_back_multiple_dlib.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@


def calculate_cdf(histogram):
print("Calculate cdf")
"""
This method calculates the cumulative distribution function
:param array histogram: The values of the histogram
Expand All @@ -40,6 +41,7 @@ def calculate_cdf(histogram):


def calculate_lookup(src_cdf, ref_cdf):
print("Calculate_lookup")
"""
This method creates the lookup table
:param array src_cdf: The cdf for the source image
Expand All @@ -60,6 +62,7 @@ def calculate_lookup(src_cdf, ref_cdf):


def match_histograms(src_image, ref_image):
print("Match histograms")
"""
This method matches the source image histogram to the
reference signal
Expand Down Expand Up @@ -125,7 +128,7 @@ def _origin_face_pts():


def compute_transformation_matrix(img, landmark, normalize, target_face_scale=1.0):

print("Compute transformation matrix")
std_pts = _standard_face_pts() # [-1,1]
target_pts = (std_pts * target_face_scale + 1) / 2 * 256.0

Expand All @@ -146,18 +149,19 @@ def compute_transformation_matrix(img, landmark, normalize, target_face_scale=1.


def compute_inverse_transformation_matrix(img, landmark, normalize, target_face_scale=1.0):
print("Compute inverse transformation matrix")

std_pts = _standard_face_pts() # [-1,1]
target_pts = (std_pts * target_face_scale + 1) / 2 * 256.0

# print(target_pts)
print(target_pts)

h, w, c = img.shape
if normalize == True:
landmark[:, 0] = landmark[:, 0] / h * 2 - 1.0
landmark[:, 1] = landmark[:, 1] / w * 2 - 1.0

# print(landmark)
print(landmark)

affine = SimilarityTransform()

Expand All @@ -167,6 +171,7 @@ def compute_inverse_transformation_matrix(img, landmark, normalize, target_face_


def show_detection(image, box, landmark):
print(" Show detection")
plt.imshow(image)
print(box[2] - box[0])
plt.gca().add_patch(
Expand All @@ -183,6 +188,7 @@ def show_detection(image, box, landmark):


def affine2theta(affine, input_w, input_h, target_w, target_h):
print("Affine2theta")
# param = np.linalg.inv(affine)
param = affine
theta = np.zeros([2, 3])
Expand All @@ -196,7 +202,7 @@ def affine2theta(affine, input_w, input_h, target_w, target_h):


def blur_blending(im1, im2, mask):

print("Blur blending")
mask *= 255.0

kernel = np.ones((10, 10), np.uint8)
Expand All @@ -215,6 +221,7 @@ def blur_blending(im1, im2, mask):


def blur_blending_cv2(im1, im2, mask):
print("Blur bending cv2")

mask *= 255.0

Expand All @@ -237,6 +244,7 @@ def blur_blending_cv2(im1, im2, mask):

# Image.composite(
def Poisson_blending(im1, im2, mask):
print("Poisson blending")

# mask=1-mask
mask *= 255
Expand All @@ -257,6 +265,7 @@ def Poisson_blending(im1, im2, mask):


def Poisson_B(im1, im2, mask, center):
print("Poisson B")

mask *= 255

Expand All @@ -268,6 +277,7 @@ def Poisson_B(im1, im2, mask, center):


def seamless_clone(old_face, new_face, raw_mask):
print("Seamless clone")

height, width, _ = old_face.shape
height = height // 2
Expand Down Expand Up @@ -310,11 +320,13 @@ def get_landmark(face_landmarks, id):
part = face_landmarks.part(id)
x = part.x
y = part.y

print("Get landmark")
print(x, y)
return (x, y)


def search(face_landmarks):
print("Search function")

x1, y1 = get_landmark(face_landmarks, 36)
x2, y2 = get_landmark(face_landmarks, 39)
Expand Down
8 changes: 6 additions & 2 deletions Face_Detection/detect_all_dlib.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,14 +82,14 @@ def compute_transformation_matrix(img, landmark, normalize, target_face_scale=1.
std_pts = _standard_face_pts() # [-1,1]
target_pts = (std_pts * target_face_scale + 1) / 2 * 256.0

# print(target_pts)
print(target_pts)

h, w, c = img.shape
if normalize == True:
landmark[:, 0] = landmark[:, 0] / h * 2 - 1.0
landmark[:, 1] = landmark[:, 1] / w * 2 - 1.0

# print(landmark)
print(landmark)

affine = SimilarityTransform()

Expand Down Expand Up @@ -171,9 +171,13 @@ def affine2theta(affine, input_w, input_h, target_w, target_h):
current_face = faces[face_id]
face_landmarks = landmark_locator(image, current_face)
current_fl = search(face_landmarks)
print("current face")
print(current_face)
plt(current_face)

affine = compute_transformation_matrix(image, current_fl, False, target_face_scale=1.3)
aligned_face = warp(image, affine, output_shape=(256, 256, 3))
plt(aligned_face)
img_name = x[:-4] + "_" + str(face_id + 1)
io.imsave(os.path.join(save_url, img_name + ".png"), img_as_ubyte(aligned_face))

Expand Down
28 changes: 14 additions & 14 deletions Global/train_domain_A.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,8 +44,8 @@
# opt.which_epoch=start_epoch-1
model = create_da_model(opt)
fd = open(path, 'w')
fd.write(str(model.module.netG))
fd.write(str(model.module.netD))
fd.write(str(model.netG))
fd.write(str(model.netD))
fd.close()

total_steps = (start_epoch - 1) * dataset_size + epoch_iter
Expand All @@ -72,7 +72,7 @@

# sum per device losses
losses = [torch.mean(x) if not isinstance(x, int) else x for x in losses]
loss_dict = dict(zip(model.module.loss_names, losses))
loss_dict = dict(zip(model.loss_names, losses))

# calculate final loss scalar
loss_D = (loss_dict['D_fake'] + loss_dict['D_real']) * 0.5
Expand All @@ -81,18 +81,18 @@

############### Backward Pass ####################
# update generator weights
model.module.optimizer_G.zero_grad()
model.optimizer_G.zero_grad()
loss_G.backward()
model.module.optimizer_G.step()
model.optimizer_G.step()

# update discriminator weights
model.module.optimizer_D.zero_grad()
model.optimizer_D.zero_grad()
loss_D.backward()
model.module.optimizer_D.step()
model.optimizer_D.step()

model.module.optimizer_featD.zero_grad()
model.optimizer_featD.zero_grad()
loss_featD.backward()
model.module.optimizer_featD.step()
model.optimizer_featD.step()

# call(["nvidia-smi", "--format=csv", "--query-gpu=memory.used,memory.free"])

Expand All @@ -101,7 +101,7 @@
if total_steps % opt.print_freq == print_delta:
errors = {k: v.data if not isinstance(v, int) else v for k, v in loss_dict.items()}
t = (time.time() - iter_start_time) / opt.batchSize
visualizer.print_current_errors(epoch, epoch_iter, errors, t, model.module.old_lr)
visualizer.print_current_errors(epoch, epoch_iter, errors, t, model.old_lr)
visualizer.plot_current_errors(errors, total_steps)

### display output images
Expand Down Expand Up @@ -133,15 +133,15 @@
### save model for this epoch
if epoch % opt.save_epoch_freq == 0:
print('saving the model at the end of epoch %d, iters %d' % (epoch, total_steps))
model.module.save('latest')
model.module.save(epoch)
model.save('latest')
model.save(epoch)
np.savetxt(iter_path, (epoch + 1, 0), delimiter=',', fmt='%d')

### instead of only training the local enhancer, train the entire network after certain iterations
if (opt.niter_fix_global != 0) and (epoch == opt.niter_fix_global):
model.module.update_fixed_params()
model.update_fixed_params()

### linearly decay learning rate after certain iterations
if epoch > opt.niter:
model.module.update_learning_rate()
model.update_learning_rate()