def _level(img, bbox, landmark, cnns, padding):
"""
LEVEL-?
"""
for i in range(5):
x, y = landmark[i]
patch, patch_bbox = getPatch(img, bbox, (x, y), padding[0])
patch = cv2.resize(patch, (15, 15)).reshape((1, 1, 15, 15))
patch = processImage(patch)
d1 = cnns[2 * i].forward(patch) # size = 1x2
patch, patch_bbox = getPatch(img, bbox, (x, y), padding[1])
patch = cv2.resize(patch, (15, 15)).reshape((1, 1, 15, 15))
patch = processImage(patch)
d2 = cnns[2 * i + 1].forward(patch)
d1 = bbox.project(patch_bbox.reproject(d1[0]))
d2 = bbox.project(patch_bbox.reproject(d2[0]))
landmark[i] = (d1 + d2) / 2
return landmark
def main(args):
inputfile = args.input
checkpoint = args.checkpoint
top_k = args.top_k
category_names = args.category_names
gpu = args.gpu
print(args)
# load checkpoint
model, optimizer = utils.loadCheckPoint(checkpoint)
# get tensor image
image = torch.from_numpy(utils.processImage(inputfile))
image.unsqueeze_(
0
) # https://discuss.pytorch.org/t/expected-stride-to-be-a-single-integer-value-or-a-list/17612/4
image = Variable(image)
# cuda is avaliable
cuda = utils.isCudaAvaliable(gpu)
if cuda:
model.cuda()
image = image.cuda()
else:
model.cpu()
# predict image
model.eval()
with torch.no_grad():
output = model.forward(image.float())
ps = torch.exp(output).data
probs, classes = ps.topk(top_k)
if cuda:
probs = probs.cpu().numpy()
classes = classes.cpu().numpy()
else:
probs = probs.numpy()
classes = classes.numpy()
classes = np.vectorize(model.idx_to_class.get)(classes)
# class name mapping
if category_names:
category_names = utils.getJSONFile(category_names)
classes = np.vectorize(category_names.get)(classes)
classes = classes[0]
probs = probs[0]
probs_max_index = np.argmax(probs)
np.set_printoptions(suppress=True)
print(classes[probs_max_index], probs[probs_max_index])
return classes[probs_max_index], probs[probs_max_index]
def level1(img, bbox, FOnly=True):
"""
LEVEL-1
img: gray image
bbox: bounding box of face
"""
F, EN, NM = getCNNs(level=1)
# F
f_bbox = bbox.subBBox(-0.05, 1.05, -0.05, 1.05)
f_face = img[f_bbox.top:f_bbox.bottom + 1, f_bbox.left:f_bbox.right + 1]
f_face = cv2.resize(f_face, (39, 39))
en_face = f_face[:31, :]
nm_face = f_face[8:, :]
f_face = f_face.reshape((1, 1, 39, 39))
f_face = processImage(f_face)
f = F.forward(f_face)
if FOnly:
return f
# EN
# en_bbox = bbox.subBBox(-0.05, 1.05, -0.04, 0.84)
# en_face = img[en_bbox.top:en_bbox.bottom+1,en_bbox.left:en_bbox.right+1]
en_face = cv2.resize(en_face, (31, 39)).reshape((1, 1, 31, 39))
en_face = processImage(en_face)
en = EN.forward(en_face)
# NM
# nm_bbox = bbox.subBBox(-0.05, 1.05, 0.18, 1.05)
# nm_face = img[nm_bbox.top:nm_bbox.bottom+1,nm_bbox.left:nm_bbox.right+1]
nm_face = cv2.resize(nm_face, (31, 39)).reshape((1, 1, 31, 39))
nm_face = processImage(nm_face)
nm = NM.forward(nm_face)
landmark = np.zeros((5, 2))
landmark[0] = (f[0] + en[0]) / 2
landmark[1] = (f[1] + en[1]) / 2
landmark[2] = (f[2] + en[2] + nm[0]) / 3
landmark[3] = (f[3] + nm[1]) / 2
landmark[4] = (f[4] + nm[2]) / 2
return landmark
def EN(img, bbox):
"""
LEVEL-1, EN
img: gray image
bbox: bounding box of face
"""
bbox = bbox.expand(0.05)
face = img[bbox.top:bbox.bottom + 1, bbox.left:bbox.right + 1]
face = cv2.resize(face, (39, 39)).reshape((1, 1, 39, 39))
face = processImage(face)
F, EN, NM = getCNNs(level=1) # TODO more flexible load needed.
landmark = EN.forward(face[:, :, :31, :])
return landmark
def EN(img, bbox):
"""
LEVEL-1, EN
img: gray image
bbox: bounding box of face
"""
bbox = bbox.expand(0.05)
face = img[bbox.top:bbox.bottom+1,bbox.left:bbox.right+1]
face = cv2.resize(face, (39, 39)).reshape((1, 1, 39, 39))
face = processImage(face)
F, EN, NM = getCNNs(level=1) # TODO more flexible load needed.
landmark = EN.forward(face[:, :, :31, :])
return landmark
def level1_Forward(img, bbox):
"""
LEVEL-1
img: gray image
bbox: bounding box of face
"""
F, _, _ = getCNNs(level=1)
# F
f_bbox = bbox.subBBox(-0.05, 1.05, -0.05, 1.05)
f_face = img[f_bbox.top:f_bbox.bottom + 1, f_bbox.left:f_bbox.right + 1]
f_face = cv2.resize(f_face, (39, 39))
f_face = f_face.reshape((1, 1, 39, 39))
f_face = processImage(f_face)
F.forward(f_face)
return F
def level1_Forward(img, bbox):
"""
LEVEL-1
img: gray image
bbox: bounding box of face
"""
F, _, _ = getCNNs(level=1)
# F
f_bbox = bbox.subBBox(-0.05, 1.05, -0.05, 1.05)
f_face = img[f_bbox.top:f_bbox.bottom+1,f_bbox.left:f_bbox.right+1]
f_face = cv2.resize(f_face, (39, 39))
f_face = f_face.reshape((1, 1, 39, 39))
f_face = processImage(f_face)
F.forward(f_face)
return F
def generate(ftxt, mode, argument=False):
"""
Generate Training Data for LEVEL-2
mode = train or test
"""
data = getDataFromTxt(ftxt)
trainData = defaultdict(lambda: dict(patches=[], landmarks=[]))
for (imgPath, bbox, landmarkGt) in data:
img = cv2.imread(imgPath, cv2.CV_LOAD_IMAGE_GRAYSCALE)
assert (img is not None)
logger("process %s" % imgPath)
landmarkPs = randomShiftWithArgument(landmarkGt, 0.05)
if not argument:
landmarkPs = [landmarkPs[0]]
for landmarkP in landmarkPs:
for idx, name, padding in types:
patch, patch_bbox = getPatch(img, bbox, landmarkP[idx],
padding)
patch = cv2.resize(patch, (15, 15))
patch = patch.reshape((1, 15, 15))
trainData[name]['patches'].append(patch)
_ = patch_bbox.project(bbox.reproject(landmarkGt[idx]))
trainData[name]['landmarks'].append(_)
for idx, name, padding in types:
logger('writing training data of %s' % name)
patches = np.asarray(trainData[name]['patches'])
landmarks = np.asarray(trainData[name]['landmarks'])
patches = processImage(patches)
shuffle_in_unison_scary(patches, landmarks)
with h5py.File('train/2_%s/%s.h5' % (name, mode), 'w') as h5:
h5['data'] = patches.astype(np.float32)
h5['landmark'] = landmarks.astype(np.float32)
with open('train/2_%s/%s.txt' % (name, mode), 'w') as fd:
fd.write('train/2_%s/%s.h5' % (name, mode))
def generate(ftxt, mode, argument=False):
"""
Generate Training Data for LEVEL-2
mode = train or test
"""
data = getDataFromTxt(ftxt)
trainData = defaultdict(lambda: dict(patches=[], landmarks=[]))
for (imgPath, bbox, landmarkGt) in data:
img = cv2.imread(imgPath, cv2.CV_LOAD_IMAGE_GRAYSCALE)
assert(img is not None)
logger("process %s" % imgPath)
landmarkPs = randomShiftWithArgument(landmarkGt, 0.05)
if not argument:
landmarkPs = [landmarkPs[0]]
for landmarkP in landmarkPs:
for idx, name, padding in types:
patch, patch_bbox = getPatch(img, bbox, landmarkP[idx], padding)
patch = cv2.resize(patch, (15, 15))
patch = patch.reshape((1, 15, 15))
trainData[name]['patches'].append(patch)
_ = patch_bbox.project(bbox.reproject(landmarkGt[idx]))
trainData[name]['landmarks'].append(_)
for idx, name, padding in types:
logger('writing training data of %s'%name)
patches = np.asarray(trainData[name]['patches'])
landmarks = np.asarray(trainData[name]['landmarks'])
patches = processImage(patches)
shuffle_in_unison_scary(patches, landmarks)
with h5py.File('train/2_%s/%s.h5'%(name, mode), 'w') as h5:
h5['data'] = patches.astype(np.float32)
h5['landmark'] = landmarks.astype(np.float32)
with open('train/2_%s/%s.txt'%(name, mode), 'w') as fd:
fd.write('train/2_%s/%s.h5'%(name, mode))
def generate_hdf5(ftxt, output, fname, argument=False):
data = getDataFromTxt(ftxt)
F_imgs = []
F_landmarks = []
for (imgPath, landmarkGt, bbox) in data:
img = cv2.imread(imgPath, cv2.IMREAD_GRAYSCALE)
assert (img is not None)
logger("process %s" % imgPath)
# plt.imshow(img)
# plt.show()
f_face = img[int(bbox[0]):int(bbox[2]), int(bbox[1]):int(bbox[3])]
plt.imshow(f_face)
plt.show()
f_face = cv2.resize(f_face, (39, 39))
f_face = f_face.reshape((1, 39, 39))
f_landmark = landmarkGt.reshape((10))
F_imgs.append(f_face)
F_landmarks.append(f_landmark)
F_imgs, F_landmarks = np.asarray(F_imgs), np.asarray(F_landmarks)
F_imgs = processImage(F_imgs)
shuffle_in_unison_scary(F_imgs, F_landmarks)
# full face
base = join(OUTPUT, '1_F')
createDir(base)
output = join(base, fname)
logger("generate %s" % output)
with h5py.File(output, 'w') as h5:
h5['data'] = F_imgs.astype(np.float32)
h5['landmark'] = F_landmarks.astype(np.float32)
###########end config###########
caffe.set_mode_gpu()
caffe.set_device(0)
model_def = 'prototxts/rpn.pt'
model_weight = 'models/vgg16_faster_reshape.caffemodel'
net = caffe.Net(model_def, model_weight, caffe.TEST)
rt = '/home/xing/py-faster-rcnn/data/drh_data/' + data_set + '/'
imgs = os.listdir(rt)
feats = {}
indx = -1
im_id = 0
for i in imgs:
indx += 1
im = cv2.imread(rt + i, 1)
try:
im = caffe.io.load_image(rt + i)
except:
continue
net.blobs['data'].reshape(1, 3, im.shape[0], im.shape[1])
pro_img = utils.processImage(im)
info = [im_id, im.shape[0], im.shape[1]]
net.blobs['info'].data[...] = np.array(info).reshape(1, 1, 1, 3)
feat = utils.getNormMac(net, pro_img)
feats[i] = feat
im_id += 1
print feat.shape
print indx, len(feat), i
np.savez('outputs/' + save_rpn_feat_name, feats=feats)
print save_rpn_feat_name, 'save done!'
def main(self):
try:
# newImgNameLst=['a','b','c','d','e']
newImgNameLst = ['2a', '2b', '2c', '2d', '2e']
# 根据生成图片及标签尺寸进行循环
for size in self.sizeLst:
tagLst = readTag(self.tagPath)
#根据所读的每个标签进行循环
for i, line in enumerate(tagLst):
if i > 25000:
return
imageInfo = line.split() #将单个数据分隔开存好
imgName = imageInfo[0]
x = int(imageInfo[1])
y = int(imageInfo[2])
# 更改默认标签框，由于原图框略大
w = int(imageInfo[3]) * 0.9
h = int(imageInfo[4]) * 0.85
if w == 0 or h == 0:
continue
x2 = x + w
y2 = y + h
#中心点坐标
centX = x + w / 2
centY = y + h / 2
#对下面操作执行多次，产生多张图片
j = 0
bigsize = 0
while j < 2:
bigsize = bigsize + 1
if bigsize > 1000: # 重复1000次，仍然没有匹配认为匹配不了
j = j + 1
#1.移动最小w，h的正负0.2倍，定义最大边长为最小w,h 0.8~最大w,h 1.2 倍
#2.按最小边倍数随机移动，得到最新框图，并且计算出坐标
movePosition1 = random.uniform(-0.2, 0.2) * min(w, h)
movePosition2 = random.uniform(-0.2, 0.2) * min(w, h)
rand = random.uniform(0.8, 1.2)
side = random.uniform(rand * min(w, h),
rand * max(w, h))
#新图形中心点坐标
newCentX = centX + movePosition1
newCentY = centY + movePosition2
newImgLeftTopX = newCentX - side / 2
newImgLeftTopY = newCentY - side / 2
newImgRightBottomX = newCentX + side / 2
newImgRightBottomY = newCentY + side / 2
offsetX1 = (newImgLeftTopX - x) / side
offsetY1 = (newImgLeftTopY - y) / side
offsetX2 = (newImgRightBottomX - x2) / side
offsetY2 = (newImgRightBottomY - y2) / side
#对原图和新图求IOU
p1 = (x, y)
p2 = (x2, y2)
newP1 = (newImgLeftTopX, newImgLeftTopY)
newP2 = (newImgRightBottomX, newImgRightBottomY)
newImgPosition = (newImgLeftTopX, newImgLeftTopY,
newImgRightBottomX,
newImgRightBottomY)
iouValue = iouFun((p1, p2, 0), (newP1, newP2, 0))
#使用三个不同值进行范围缩放
#分别执行，1.从原图抠图 2.保存不同尺寸图片 3.保存坐标文件
imgPath2 = ''
confidence = 0
if iouValue > 0.7:
imgPath2 = 'positive'
confidence = MyEnum.positive.value
elif iouValue < 0.3:
imgPath2 = 'negative'
confidence = MyEnum.negative.value
# elif iouValue>0.1 and iouValue<0.35:
# if iouValue>=0 and iouValue<0.35:
# imgPath2='part'
# confidence=MyEnum.part.value
if imgPath2:
newImgName = newImgNameLst[j] + imgName
offset = (newImgName, confidence, offsetX1,
offsetY1, offsetX2, offsetY2)
j = j + 1
print("第{}轮，第{}次".format(i, j))
processImage(newImgName,
imgName,
self.imagePath,
self.saveImgPath,
imgPath2,
self.saveTagPath,
offset,
newImgPosition,
outImgSize=size)
print("Done...............", size)
except Exception as e:
print("ERROR:", "main" + str(e))
def generate_hdf5_data(filelist, output, fname, argument=False):
data = getDataFromTxt(filelist)
F_imgs = []
F_landmarks = []
EN_imgs = []
EN_landmarks = []
NM_imgs = []
NM_landmarks = []
for (imgPath, bbox, landmarkGt) in data:
img = cv2.imread(imgPath, cv2.CV_LOAD_IMAGE_GRAYSCALE)
assert (img is not None)
logger("process %s" % imgPath)
# Paper Table2 jitter F-layer
f_bbox = bbox.subBBox(-0.05, 1.05, -0.05, 1.05)
f_face = img[f_bbox.top:f_bbox.bottom + 1,
f_bbox.left:f_bbox.right + 1]
## data argument
if argument and np.random.rand() > -1:
### flip
face_flipped, landmark_flipped = flip(f_face, landmarkGt)
face_flipped = cv2.resize(face_flipped, (39, 39))
F_imgs.append(face_flipped.reshape((1, 39, 39)))
F_landmarks.append(landmark_flipped.reshape(10))
### rotation +5 degrees
if np.random.rand() > 0.5:
face_rotated_by_alpha, landmark_rotated = rotate(img, f_bbox, \
bbox.reprojectLandmark(landmarkGt), 5)
landmark_rotated = bbox.projectLandmark(landmark_rotated)
face_rotated_by_alpha = cv2.resize(face_rotated_by_alpha,
(39, 39))
F_imgs.append(face_rotated_by_alpha.reshape((1, 39, 39)))
F_landmarks.append(landmark_rotated.reshape(10))
### flip with rotation
face_flipped, landmark_flipped = flip(face_rotated_by_alpha,
landmark_rotated)
face_flipped = cv2.resize(face_flipped, (39, 39))
F_imgs.append(face_flipped.reshape((1, 39, 39)))
F_landmarks.append(landmark_flipped.reshape(10))
### rotation -5 degrees
if np.random.rand() > 0.5:
face_rotated_by_alpha, landmark_rotated = rotate(img, f_bbox, \
bbox.reprojectLandmark(landmarkGt), -5)
landmark_rotated = bbox.projectLandmark(landmark_rotated)
face_rotated_by_alpha = cv2.resize(face_rotated_by_alpha,
(39, 39))
F_imgs.append(face_rotated_by_alpha.reshape((1, 39, 39)))
F_landmarks.append(landmark_rotated.reshape(10))
### flip with rotation
face_flipped, landmark_flipped = flip(face_rotated_by_alpha,
landmark_rotated)
face_flipped = cv2.resize(face_flipped, (39, 39))
F_imgs.append(face_flipped.reshape((1, 39, 39)))
F_landmarks.append(landmark_flipped.reshape(10))
f_face = cv2.resize(f_face, (39, 39))
en_face = f_face[:31, :]
nm_face = f_face[8:, :]
f_face = f_face.reshape((1, 39, 39))
f_landmark = landmarkGt.reshape((10))
F_imgs.append(f_face)
F_landmarks.append(f_landmark)
## data argument for EN
if argument and np.random.rand() > 0.5:
### flip
face_flipped, landmark_flipped = flip(en_face, landmarkGt)
face_flipped = cv2.resize(face_flipped, (31, 39)).reshape(
(1, 31, 39))
landmark_flipped = landmark_flipped[:3, :].reshape((6))
EN_imgs.append(face_flipped)
EN_landmarks.append(landmark_flipped)
en_face = cv2.resize(en_face, (31, 39)).reshape((1, 31, 39))
en_landmark = landmarkGt[:3, :].reshape((6))
EN_imgs.append(en_face)
EN_landmarks.append(en_landmark)
## data argument for NM
if argument and np.random.rand() > 0.5:
### flip
face_flipped, landmark_flipped = flip(nm_face, landmarkGt)
face_flipped = cv2.resize(face_flipped, (31, 39)).reshape(
(1, 31, 39))
landmark_flipped = landmark_flipped[2:, :].reshape((6))
NM_imgs.append(face_flipped)
NM_landmarks.append(landmark_flipped)
nm_face = cv2.resize(nm_face, (31, 39)).reshape((1, 31, 39))
nm_landmark = landmarkGt[2:, :].reshape((6))
NM_imgs.append(nm_face)
NM_landmarks.append(nm_landmark)
#Convert the list to array
F_imgs, F_landmarks = np.asarray(F_imgs), np.asarray(F_landmarks)
EN_imgs, EN_landmarks = np.asarray(EN_imgs), np.asarray(EN_landmarks)
NM_imgs, NM_landmarks = np.asarray(NM_imgs), np.asarray(NM_landmarks)
### normalize the data and shu
F_imgs = processImage(F_imgs)
shuffle_in_unison_scary(F_imgs, F_landmarks)
EN_imgs = processImage(EN_imgs)
shuffle_in_unison_scary(EN_imgs, EN_landmarks)
NM_imgs = processImage(NM_imgs)
shuffle_in_unison_scary(NM_imgs, NM_landmarks)
# full face
base = join(OUTPUT, '1_F')
createDir(base)
output = join(base, fname)
logger("generate %s" % output)
with h5py.File(output, 'w') as h5:
h5['data'] = F_imgs.astype(np.float32)
h5['landmark'] = F_landmarks.astype(np.float32)
# eye and nose
base = join(OUTPUT, '1_EN')
createDir(base)
output = join(base, fname)
logger("generate %s" % output)
with h5py.File(output, 'w') as h5:
h5['data'] = EN_imgs.astype(np.float32)
h5['landmark'] = EN_landmarks.astype(np.float32)
# nose and mouth
base = join(OUTPUT, '1_NM')
createDir(base)
output = join(base, fname)
logger("generate %s" % output)
with h5py.File(output, 'w') as h5:
h5['data'] = NM_imgs.astype(np.float32)
h5['landmark'] = NM_landmarks.astype(np.float32)
def generate_hdf5_data(filelist, output, fname, argument=False):
data = getDataFromTxt(filelist)
F_imgs = []
F_landmarks = []
EN_imgs = []
EN_landmarks = []
NM_imgs = []
NM_landmarks = []
for (imgPath, bbox, landmarkGt) in data:
img = cv2.imread(imgPath, cv2.CV_LOAD_IMAGE_GRAYSCALE)
assert(img is not None)
logger("process %s" % imgPath)
# Paper Table2 jitter F-layer
f_bbox = bbox.subBBox(-0.05, 1.05, -0.05, 1.05)
f_face = img[f_bbox.top:f_bbox.bottom+1,f_bbox.left:f_bbox.right+1]
## data argument
if argument and np.random.rand() > -1:
### flip
face_flipped, landmark_flipped = flip(f_face, landmarkGt)
face_flipped = cv2.resize(face_flipped, (39, 39))
F_imgs.append(face_flipped.reshape((1, 39, 39)))
F_landmarks.append(landmark_flipped.reshape(10))
### rotation +5 degrees
if np.random.rand() > 0.5:
face_rotated_by_alpha, landmark_rotated = rotate(img, f_bbox, \
bbox.reprojectLandmark(landmarkGt), 5)
landmark_rotated = bbox.projectLandmark(landmark_rotated)
face_rotated_by_alpha = cv2.resize(face_rotated_by_alpha, (39, 39))
F_imgs.append(face_rotated_by_alpha.reshape((1, 39, 39)))
F_landmarks.append(landmark_rotated.reshape(10))
### flip with rotation
face_flipped, landmark_flipped = flip(face_rotated_by_alpha, landmark_rotated)
face_flipped = cv2.resize(face_flipped, (39, 39))
F_imgs.append(face_flipped.reshape((1, 39, 39)))
F_landmarks.append(landmark_flipped.reshape(10))
### rotation -5 degrees
if np.random.rand() > 0.5:
face_rotated_by_alpha, landmark_rotated = rotate(img, f_bbox, \
bbox.reprojectLandmark(landmarkGt), -5)
landmark_rotated = bbox.projectLandmark(landmark_rotated)
face_rotated_by_alpha = cv2.resize(face_rotated_by_alpha, (39, 39))
F_imgs.append(face_rotated_by_alpha.reshape((1, 39, 39)))
F_landmarks.append(landmark_rotated.reshape(10))
### flip with rotation
face_flipped, landmark_flipped = flip(face_rotated_by_alpha, landmark_rotated)
face_flipped = cv2.resize(face_flipped, (39, 39))
F_imgs.append(face_flipped.reshape((1, 39, 39)))
F_landmarks.append(landmark_flipped.reshape(10))
f_face = cv2.resize(f_face, (39, 39))
en_face = f_face[:31, :]
nm_face = f_face[8:, :]
f_face = f_face.reshape((1, 39, 39))
f_landmark = landmarkGt.reshape((10))
F_imgs.append(f_face)
F_landmarks.append(f_landmark)
## data argument for EN
if argument and np.random.rand() > 0.5:
### flip
face_flipped, landmark_flipped = flip(en_face, landmarkGt)
face_flipped = cv2.resize(face_flipped, (31, 39)).reshape((1, 31, 39))
landmark_flipped = landmark_flipped[:3, :].reshape((6))
EN_imgs.append(face_flipped)
EN_landmarks.append(landmark_flipped)
en_face = cv2.resize(en_face, (31, 39)).reshape((1, 31, 39))
en_landmark = landmarkGt[:3, :].reshape((6))
EN_imgs.append(en_face)
EN_landmarks.append(en_landmark)
## data argument for NM
if argument and np.random.rand() > 0.5:
### flip
face_flipped, landmark_flipped = flip(nm_face, landmarkGt)
face_flipped = cv2.resize(face_flipped, (31, 39)).reshape((1, 31, 39))
landmark_flipped = landmark_flipped[2:, :].reshape((6))
NM_imgs.append(face_flipped)
NM_landmarks.append(landmark_flipped)
nm_face = cv2.resize(nm_face, (31, 39)).reshape((1, 31, 39))
nm_landmark = landmarkGt[2:, :].reshape((6))
NM_imgs.append(nm_face)
NM_landmarks.append(nm_landmark)
#Convert the list to array
F_imgs, F_landmarks = np.asarray(F_imgs), np.asarray(F_landmarks)
EN_imgs, EN_landmarks = np.asarray(EN_imgs), np.asarray(EN_landmarks)
NM_imgs, NM_landmarks = np.asarray(NM_imgs),np.asarray(NM_landmarks)
### normalize the data and shu
F_imgs = processImage(F_imgs)
shuffle_in_unison_scary(F_imgs, F_landmarks)
EN_imgs = processImage(EN_imgs)
shuffle_in_unison_scary(EN_imgs, EN_landmarks)
NM_imgs = processImage(NM_imgs)
shuffle_in_unison_scary(NM_imgs, NM_landmarks)
# full face
base = join(OUTPUT, '1_F')
createDir(base)
output = join(base, fname)
logger("generate %s" % output)
with h5py.File(output, 'w') as h5:
h5['data'] = F_imgs.astype(np.float32)
h5['landmark'] = F_landmarks.astype(np.float32)
# eye and nose
base = join(OUTPUT, '1_EN')
createDir(base)
output = join(base, fname)
logger("generate %s" % output)
with h5py.File(output, 'w') as h5:
h5['data'] = EN_imgs.astype(np.float32)
h5['landmark'] = EN_landmarks.astype(np.float32)
# nose and mouth
base = join(OUTPUT, '1_NM')
createDir(base)
output = join(base, fname)
logger("generate %s" % output)
with h5py.File(output, 'w') as h5:
h5['data'] = NM_imgs.astype(np.float32)
h5['landmark'] = NM_landmarks.astype(np.float32)
#分别执行，1.从原图抠图 2.保存不同尺寸图片 3.保存坐标文件
imgPath2=''
confidence=0
# if iouValue>0.65:
# imgPath2='positive'
# confidence=MyEnum.positive.value
# el
# if iouValue<0.3:
# imgPath2='negative'
# confidence=MyEnum.negative.value
# el
if iouValue>0.4 or iouValue<0.65:
imgPath2='part'
confidence=MyEnum.part.value
if imgPath2:
newImgName='b'+"-"+str(s)+"-"+str(j)+"-"+image_name_detail
if imgPath2=='negative':
offset=(newImgName,confidence,0,0,0,0)
else:
offset=(newImgName,confidence,offset_x1,offset_y1,offset_x2,offset_y2)
j=j+1
print("生成{}尺寸，第{}轮，第{}次".format(my_format,i,j))
processImage(newImgName,image_name_detail,image_name,save_images_folder,imgPath2,save_tag_path,offset,newImgPosition,outImgSize=my_format)
except Exception as e:
print("ERROR:","__name__"+str(e))
# cv2.rectangle(image, pt1=(x, y), pt2=(x + w, y + h),color=(255,0,0))
# for i, sample in enumerate(dataset):
# print(i, sample['image'])
#
# print(len(dataset))