github/DeepFaceLab
1
0
Fork 0
mirror of https://github.com/iperov/DeepFaceLab.git synced 2025-07-11 15:47:01 -07:00
Code Issues Projects Releases Packages Wiki Activity Actions

update SampleGeneratorFaceSkinSegDataset

Browse source
This commit is contained in:
Colombo 2020-03-13 19:27:27 +04:00
parent 7c89077321
commit 144675020c
1 changed files with 80 additions and 79 deletions
Download patch file Download diff file Expand all files Collapse all files

81
samplelib/SampleGeneratorFaceSkinSegDataset.py
View file

@ -8,6 +8,7 @@ import cv2
import numpy as np import numpy as np
from core import imagelib, mplib, pathex from core import imagelib, mplib, pathex
from core.imagelib import sd
from core.cv2ex import * from core.cv2ex import *
from core.interact import interact as io from core.interact import interact as io
from core.joblib import SubprocessGenerator, ThisThreadGenerator from core.joblib import SubprocessGenerator, ThisThreadGenerator
@ -121,8 +122,9 @@ class SampleGeneratorFaceSkinSegDataset(SampleGeneratorBase):
samples = pickle.loads(pickled_samples) samples = pickle.loads(pickled_samples)
obstructions_images_paths_len = len(obstructions_images_paths)
shuffle_o_idxs = [] shuffle_o_idxs = []
o_idxs = [*range(len(obstructions_images_paths))] o_idxs = [*range(obstructions_images_paths_len)]
shuffle_idxs = [] shuffle_idxs = []
idxs = [*range(len(samples))] idxs = [*range(len(samples))]
@ -178,58 +180,59 @@ class SampleGeneratorFaceSkinSegDataset(SampleGeneratorBase):
if len(mask.shape) == 2: if len(mask.shape) == 2:
mask = mask[...,None] mask = mask[...,None]
# apply obstruction if obstructions_images_paths_len != 0:
if len(shuffle_o_idxs) == 0: # apply obstruction
shuffle_o_idxs = o_idxs.copy() if len(shuffle_o_idxs) == 0:
np.random.shuffle(shuffle_o_idxs) shuffle_o_idxs = o_idxs.copy()
o_idx = shuffle_o_idxs.pop() np.random.shuffle(shuffle_o_idxs)
o_img = cv2_imread (obstructions_images_paths[o_idx]).astype(np.float32) / 255.0 o_idx = shuffle_o_idxs.pop()
oh,ow,oc = o_img.shape o_img = cv2_imread (obstructions_images_paths[o_idx]).astype(np.float32) / 255.0
if oc == 4: oh,ow,oc = o_img.shape
ohw = max(oh,ow) if oc == 4:
scale = resolution / ohw ohw = max(oh,ow)
scale = resolution / ohw
#o_img = cv2.resize (o_img, ( int(ow*rate), int(oh*rate), ), cv2.INTER_CUBIC) #o_img = cv2.resize (o_img, ( int(ow*rate), int(oh*rate), ), cv2.INTER_CUBIC)
mat = cv2.getRotationMatrix2D( (ow/2,oh/2), mat = cv2.getRotationMatrix2D( (ow/2,oh/2),
np.random.uniform( o_rotation_range[0], o_rotation_range[1] ), np.random.uniform( o_rotation_range[0], o_rotation_range[1] ),
1.0 ) 1.0 )
mat += np.float32( [[0,0, -ow/2 ], mat += np.float32( [[0,0, -ow/2 ],
[0,0, -oh/2 ]]) [0,0, -oh/2 ]])
mat *= scale * np.random.uniform(1 +o_scale_range[0], 1 +o_scale_range[1]) mat *= scale * np.random.uniform(1 +o_scale_range[0], 1 +o_scale_range[1])
mat += np.float32( [[0, 0, resolution/2 + resolution*np.random.uniform( o_tx_range[0], o_tx_range[1] ) ], mat += np.float32( [[0, 0, resolution/2 + resolution*np.random.uniform( o_tx_range[0], o_tx_range[1] ) ],
[0, 0, resolution/2 + resolution*np.random.uniform( o_ty_range[0], o_ty_range[1] ) ] ]) [0, 0, resolution/2 + resolution*np.random.uniform( o_ty_range[0], o_ty_range[1] ) ] ])
o_img = cv2.warpAffine( o_img, mat, (resolution,resolution), borderMode=cv2.BORDER_CONSTANT, flags=cv2.INTER_LANCZOS4 ) o_img = cv2.warpAffine( o_img, mat, (resolution,resolution), borderMode=cv2.BORDER_CONSTANT, flags=cv2.INTER_LANCZOS4 )
if o_random_flip and np.random.randint(10) < 4: if o_random_flip and np.random.randint(10) < 4:
o_img = o_img[:,::-1,...] o_img = o_img[:,::-1,...]
o_mask = o_img[...,3:4] o_mask = o_img[...,3:4]
o_mask[o_mask>0] = 1.0 o_mask[o_mask>0] = 1.0
o_mask = cv2.erode (o_mask, cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(5,5)), iterations = 1 ) o_mask = cv2.erode (o_mask, cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(5,5)), iterations = 1 )
o_mask = cv2.GaussianBlur(o_mask, (5, 5) , 0)[...,None] o_mask = cv2.GaussianBlur(o_mask, (5, 5) , 0)[...,None]
img = img*(1-o_mask) + o_img[...,0:3]*o_mask img = img*(1-o_mask) + o_img[...,0:3]*o_mask
o_mask[o_mask<0.5] = 0.0 o_mask[o_mask<0.5] = 0.0
#import code #import code
#code.interact(local=dict(globals(), **locals())) #code.interact(local=dict(globals(), **locals()))
mask *= (1-o_mask) mask *= (1-o_mask)
#cv2.imshow ("", np.clip(o_img*255, 0,255).astype(np.uint8) ) #cv2.imshow ("", np.clip(o_img*255, 0,255).astype(np.uint8) )
#cv2.waitKey(0) #cv2.waitKey(0)
img = imagelib.warp_by_params (warp_params, img, can_warp=True, can_transform=True, can_flip=True, border_replicate=False) img = imagelib.warp_by_params (warp_params, img, can_warp=True, can_transform=True, can_flip=True, border_replicate=False)
@ -241,13 +244,11 @@ class SampleGeneratorFaceSkinSegDataset(SampleGeneratorBase):
mask[mask >= 0.5] = 1.0 mask[mask >= 0.5] = 1.0
mask = np.clip(mask, 0, 1) mask = np.clip(mask, 0, 1)
img = imagelib.apply_random_hsv_shift(img)
#todo random mask for blur img = imagelib.apply_random_hsv_shift(img, mask=sd.random_circle_faded ([resolution,resolution]))
img = imagelib.apply_random_motion_blur( img, motion_blur_chance, motion_blur_mb_max_size, mask=sd.random_circle_faded ([resolution,resolution]))
img = imagelib.apply_random_motion_blur( img, motion_blur_chance, motion_blur_mb_max_size ) img = imagelib.apply_random_gaussian_blur( img, gaussian_blur_chance, gaussian_blur_kernel_max_size, mask=sd.random_circle_faded ([resolution,resolution]))
img = imagelib.apply_random_gaussian_blur( img, gaussian_blur_chance, gaussian_blur_kernel_max_size ) img = imagelib.apply_random_bilinear_resize( img, random_bilinear_resize_chance, random_bilinear_resize_max_size_per, mask=sd.random_circle_faded ([resolution,resolution]))
img = imagelib.apply_random_bilinear_resize( img, random_bilinear_resize_chance, random_bilinear_resize_max_size_per )
if data_format == "NCHW": if data_format == "NCHW":
img = np.transpose(img, (2,0,1) ) img = np.transpose(img, (2,0,1) )