def test_preserve_aspect_ratio():
img_path = os.path.join(here, 'data', 'green_fish.jpg')
img = caer.imread(img_path)
img_400_400 = caer.resize(img, target_size=(400,400), preserve_aspect_ratio=True)
img_223_182 = caer.resize(img, target_size=(223,182), preserve_aspect_ratio=True)
img_93_35 = caer.resize(img, target_size=(93,35), preserve_aspect_ratio=True)
assert img_400_400.shape[:2] == (400,400)
assert img_223_182.shape[:2] == (182,223) # Numpy arrays are processed differently (h,w) as opposed to (w,h)
assert img_93_35.shape[:2] == (35,93) # Numpy arrays are processed differently (h,w) as opposed to (w,h)
def test_target_sizes():
img_path = os.path.join(here, 'data', 'beverages.jpg')
img = caer.imread(img_path)
img_400_400 = caer.resize(img, target_size=(400,400))
img_304_339 = caer.resize(img, target_size=(304,339))
img_199_206 = caer.resize(img, target_size=(199,206))
assert img_400_400.shape[:2] == (400,400)
assert img_304_339.shape[:2] == (339,304) # Numpy arrays are processed differently (h,w) as opposed to (w,h)
assert img_199_206.shape[:2] == (206,199) # Numpy arrays are processed differently (h,w) as opposed to (w,h)
def test_preserve_aspect_ratio():
tens_400_400 = caer.resize(cv_tens, target_size=(400,400), preserve_aspect_ratio=True)
tens_223_182 = caer.resize(cv_tens, target_size=(223,182), preserve_aspect_ratio=True)
tens_93_35 = caer.resize(cv_tens, target_size=(93,35), preserve_aspect_ratio=True)
assert tens_400_400.shape[:2] == (400,400)
assert tens_223_182.shape[:2] == (182,223)
assert tens_93_35.shape[:2] == (35,93)
# Type Asserts
## Using isinstance() often mistakes a caer.Tensor as an np.ndarray
assert 'caer.Tensor' in str(type(tens_400_400))
assert 'caer.Tensor' in str(type(tens_223_182))
assert 'caer.Tensor' in str(type(tens_93_35))
def show_resized_image():
global currentImage
global resizedImgBtn
global flipHImgBtn
global flipVImgBtn
global flipHVImgBtn
global rotateImgBtn
tempSize = selectedSize.get()
if 'x' in tempSize:
size = tempSize.replace(' ', '').split('x')
try:
if resizedImgBtn['bg'] == 'lightgrey':
resizedImgBtn['bg'] = 'lightblue'
if flipHImgBtn['bg'] == 'lightblue':
flipHImgBtn['bg'] = 'lightgrey'
elif flipVImgBtn['bg'] == 'lightblue':
flipVImgBtn['bg'] = 'lightgrey'
elif flipHVImgBtn['bg'] == 'lightblue':
flipHVImgBtn['bg'] = 'lightgrey'
else:
rotateImgBtn['bg'] = 'lightgrey'
if not transformedImage is None:
currentImage = transformedImage
reset_ghs()
# Resize the image without preserving aspect ratio
currentImage = caer.resize(currentImage,
target_size=(int(size[0]),
int(size[1])),
preserve_aspect_ratio=False)
currentImage.cspace = 'rgb'
if rotationApplied:
show_rotated_image()
else:
image_show(currentImage)
except Exception as e:
print(str(e))
def test_target_sizes():
# Should return <ndarray>s
tens_400_400 = caer.resize(cv_tens, target_size=(400, 400))
tens_304_339 = caer.resize(cv_tens, target_size=(304, 339))
tens_199_206 = caer.resize(cv_tens, target_size=(199, 206))
# Should return <caer.Tensor>s
caer_tens_400_400 = caer.resize(caer_tens, target_size=(400, 400))
caer_tens_304_339 = caer.resize(caer_tens, target_size=(304, 339))
caer_tens_199_206 = caer.resize(caer_tens, target_size=(199, 206))
assert tens_400_400.shape[:2] == (400, 400)
assert tens_304_339.shape[:2] == (339, 304)
assert tens_199_206.shape[:2] == (206, 199)
# Type Asserts
## Using isinstance() often mistakes a caer.Tensor as an np.ndarray
assert 'caer.Tensor' in str(type(tens_400_400))
assert 'caer.Tensor' in str(type(tens_304_339))
assert 'caer.Tensor' in str(type(tens_199_206))
def adjust_ghsps(*args):
global transformedImage
if not currentImage is None:
# reset the error label's text
if lblError['text'] == 'Error':
lblError['text'] = ''
transformedImage = caer.to_tensor(currentImage, cspace="rgb")
# apply all transformations to currently displayed image
if image_resized:
transformedImage = caer.resize(transformedImage,
target_size=(int(image_size[0]),
int(image_size[1])),
preserve_aspect_ratio=False)
if hue.get() != 0.0:
transformedImage = caer.transforms.adjust_hue(
transformedImage, hue.get())
if saturation.get() != 1.0:
transformedImage = caer.transforms.adjust_saturation(
transformedImage, saturation.get())
if imgGamma.get() != 1.05:
transformedImage = caer.transforms.adjust_gamma(
transformedImage, imgGamma.get())
if sharpen.get() != 8.9:
transformedImage = caer.core.cv.filter2D(transformedImage, -1,
sharpenKernel)
gb = gaussian_blur.get()
if gb > 1:
transformedImage = caer.core.cv.GaussianBlur(
transformedImage, (gb + 1, gb + 1),
caer.core.cv.BORDER_DEFAULT)
if posterize.get() < 6:
transformedImage = caer.transforms.posterize(
transformedImage, posterize.get())
if solarize.get() < 255:
transformedImage = caer.transforms.solarize(
transformedImage, solarize.get())
if sobel_threshold.get() > 0:
transformedImage = caer.core.cv.cvtColor(
transformedImage, caer.core.cv.COLOR_RGB2GRAY)
sobelKernel = sobel_threshold.get() if sobel_threshold.get(
) % 2 != 0 else sobel_threshold.get() + 1 # values 1, 3 and 5
dx = dy = sobel_threshold.get() - 2 if sobel_threshold.get(
) > 2 else sobel_threshold.get()
sobelx = caer.core.cv.Sobel(transformedImage,
caer.core.cv.IMREAD_GRAYSCALE,
dx,
0,
ksize=sobelKernel)
sobely = caer.core.cv.Sobel(transformedImage,
caer.core.cv.IMREAD_GRAYSCALE,
0,
dy,
ksize=sobelKernel)
transformedImage = caer.core.cv.bitwise_or(sobelx, sobely)
transformedImage = caer.core.cv.cvtColor(
transformedImage, caer.core.cv.COLOR_GRAY2RGB)
if show_edges.get() == 1:
transformedImage = caer.core.cv.cvtColor(
transformedImage, caer.core.cv.COLOR_RGB2GRAY)
transformedImage = caer.core.cv.Canny(transformedImage,
low_threshold.get(),
low_threshold.get() * 2)
transformedImage = caer.core.cv.cvtColor(
transformedImage, caer.core.cv.COLOR_GRAY2RGB)
if show_emboss.get() == 1:
transformedImage = caer.core.cv.filter2D(
transformedImage, -1, embossKernel) + emboss.get()
if flip_H:
transformedImage = caer.transforms.hflip(transformedImage)
if flip_V:
transformedImage = caer.transforms.vflip(transformedImage)
if rotationApplied:
show_rotated_image(True)
else:
image_show(transformedImage)
import cv2 as cv
import caer
img = cv.imread('./dataset/testing_images/bryan-1.jpg')
resized = caer.resize(img, target_size=(400, 400), preserve_aspect_ratio=True)
cv.imshow('Detected Faces', img)
gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY)
cv.imshow('Gray People', gray)
haar_cascade = cv.CascadeClassifier('./cascade/cascade.xml')
faces_rect = haar_cascade.detectMultiScale(gray,
scaleFactor=1.1,
minNeighbors=1)
print(f'Number of faces found = {len(faces_rect)}')
for (x, y, w, h) in faces_rect:
cv.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), thickness=2)
cv.imshow('Detected Faces', img)
cv.waitKey(0)
