from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import os
import imageio
import scipy.misc as misc
import scipy
import numpy as np
import cv2
def save_as_gif(images_list, out_path, gif_file_name='all', save_image=False):
if os.path.exists(out_path) == False:
os.mkdir(out_path)
# save as .png
if save_image == True:
for n in range(len(images_list)):
file_name = '{}.png'.format(n)
save_path_and_name = os.path.join(out_path, file_name)
misc.imsave(save_path_and_name, images_list[n])
# save as .gif
out_path_and_name = os.path.join(out_path, '{}.gif'.format(gif_file_name))
imageio.mimsave(out_path_and_name, images_list, 'GIF', duration=0.1)
def get_image(image_path, crop_size=128, is_crop=False, resize_w=140, is_grayscale=False):
return transform(imread(image_path , is_grayscale), crop_size, is_crop, resize_w)
def transform(image, crop_size=64, is_crop=True, resize_w=140):
image = scipy.misc.imresize(image, [resize_w, resize_w])
if is_crop:
cropped_image = center_crop(image, crop_size)
else:
cropped_image = image
cropped_image = scipy.misc.imresize(cropped_image ,
[resize_w , resize_w])
return np.array(cropped_image) / 127.5 - 1
def center_crop(x, crop_h, crop_w=None):
if crop_w is None:
crop_w = crop_h
h, w = x.shape[:2]
j = int(round((h - crop_h)/2.))
i = int(round((w - crop_w)/2.))
rate = np.random.uniform(0, 1, size=1)
if rate < 0.5:
x = np.fliplr(x)
return x[j:j+crop_h, i:i+crop_w]
def transform_image(image):
return (image + 1) * 127.5
def save_images(images, image_path, is_verse=True):
if is_verse:
return imageio.imwrite(image_path, inverse_transform(images))
else:
return imageio.imwrite(image_path, images)
def resizeImg(img, size=list):
return scipy.misc.imresize(img, size)
def imread(path, is_grayscale=False):
if (is_grayscale):
return scipy.misc.imread(path, flatten=True).astype(np.float)
else:
return scipy.misc.imread(path).astype(np.float)
def imsave(images, size, path):
return scipy.misc.imsave(path, merge(images, size))
def merge(images, size):
if len(images.shape) == 3:
img = images
else:
h, w = images.shape[1], images.shape[2]
img = np.zeros((int(h * size[0]), int(w * size[1]), 3))
for idx, image in enumerate(images):
i = idx % size[1]
j = idx // size[1]
img[j * h:j * h + h, i * w: i * w + w, :] = image
return img
def inverse_transform(image):
result = ((image + 1) * 127.5).astype(np.uint8)
return result
height_to_eyeball_radius_ratio = 1.1
eyeball_radius_to_iris_diameter_ratio = 1.0
def from_gaze2d(gaze, output_size, scale=1.0):
"""Generate a normalized pictorial representation of 3D gaze direction."""
gazemaps = []
oh, ow = np.round(scale * np.asarray(output_size)).astype(np.int32)
oh_2 = int(np.round(0.5 * oh))
ow_2 = int(np.round(0.5 * ow))
r = int(height_to_eyeball_radius_ratio * oh_2)
theta, phi = gaze
theta = -theta
sin_theta = np.sin(theta)
cos_theta = np.cos(theta)
sin_phi = np.sin(phi)
cos_phi = np.cos(phi)
# Draw iris
eyeball_radius = int(height_to_eyeball_radius_ratio * oh_2)
iris_radius_angle = np.arcsin(0.5 * eyeball_radius_to_iris_diameter_ratio)
iris_radius = eyeball_radius_to_iris_diameter_ratio * eyeball_radius
iris_distance = float(eyeball_radius) * np.cos(iris_radius_angle)
iris_offset = np.asarray([
-iris_distance * sin_phi * cos_theta,
iris_distance * sin_theta,
])
iris_centre = np.asarray([ow_2, oh_2]) + iris_offset
angle = np.degrees(np.arctan2(iris_offset[1], iris_offset[0]))
ellipse_max = eyeball_radius_to_iris_diameter_ratio * iris_radius
ellipse_min = np.abs(ellipse_max * cos_phi * cos_theta)
#gazemap = np.zeros((oh, ow), dtype=np.float32)
# Draw eyeball
gazemap = np.zeros((oh, ow), dtype=np.float32)
gazemap = cv2.ellipse(gazemap, box=(iris_centre, (ellipse_min, ellipse_max), angle),
color = 1.0 , thickness=-1, lineType=cv2.LINE_AA)
#outout = cv2.circle(test_gazemap, (ow_2, oh_2), r, color=1, thickness=-1)
gazemaps.append(gazemap)
gazemap = np.zeros((oh, ow), dtype=np.float32)
gazemap = cv2.circle(gazemap, (ow_2, oh_2), r, color=1, thickness=-1)
gazemaps.append(gazemap)
return np.asarray(gazemaps)
if __name__ == "__main__":
target_angles = [[0.0, -1.0], [1.0, -1.0], [1.0, -0.66], [1.0, -0.33], [1.0, 0.0], [1.0, 0.33],
[1.0, 0.66], [1.0, 1.0], [0, 1.0], [-1.0, 1.0]]
for i, angles in enumerate(target_angles):
x, y = angles[0], - 1 * angles[1]
gazemaps = from_gaze2d((x, y), (64, 64))
cv2.imwrite("gazemaps_{}.jpg".format(i), gazemaps[1,...] * 255.0)
