def _read(self, idx): image = datasets.imread_rgb(os.path.join(self._root, "Images", self._data[idx])) if self._crop != False: imheight, imwidth = image.shape[:2] annotation = ET.parse(os.path.join(self._root, "Annotation", self._data[idx][:-4])).getroot() objects = [c for c in annotation if c.tag == "object"] bbox = objects[0].find('bndbox') xmin = int(bbox.find('xmin').text) - 1 ymin = int(bbox.find('ymin').text) - 1 xmax = int(bbox.find('xmax').text) ymax = int(bbox.find('ymax').text) # expand width = xmax - xmin height = ymax - ymin centerx = xmin + width / 2. centery = ymin + height / 2. xoffset = width * self._crop / 2. yoffset = height * self._crop / 2. xmin = max(int(centerx - xoffset + 0.5), 0) ymin = max(int(centery - yoffset + 0.5), 0) xmax = min(int(centerx + xoffset + 0.5), imwidth - 1) ymax = min(int(centery + yoffset + 0.5), imheight - 1) image = image[ymin:ymax, xmin:xmax].copy() if self._target_size is not None: image = misc.imresize(image, self._target_size) return image
def _read(self, idx): image = datasets.imread_rgb( os.path.join(self._root, "Images", self._data[idx])) if self._crop != False: imheight, imwidth = image.shape[:2] annotation = ET.parse( os.path.join(self._root, "Annotation", self._data[idx][:-4])).getroot() objects = [c for c in annotation if c.tag == "object"] bbox = objects[0].find('bndbox') xmin = int(bbox.find('xmin').text) - 1 ymin = int(bbox.find('ymin').text) - 1 xmax = int(bbox.find('xmax').text) ymax = int(bbox.find('ymax').text) # expand width = xmax - xmin height = ymax - ymin centerx = xmin + width / 2. centery = ymin + height / 2. xoffset = width * self._crop / 2. yoffset = height * self._crop / 2. xmin = max(int(centerx - xoffset + 0.5), 0) ymin = max(int(centery - yoffset + 0.5), 0) xmax = min(int(centerx + xoffset + 0.5), imwidth - 1) ymax = min(int(centery + yoffset + 0.5), imheight - 1) image = image[ymin:ymax, xmin:xmax].copy() if self._target_size is not None: image = misc.imresize(image, self._target_size) return image
def _read(self, idx): image = datasets.imread_rgb(self._data[idx]) self._raw_dimension[idx] = image.shape[:2] xmin, ymin, xmax, ymax = \ self._get_cropped_coordinates(idx) image = image[ymin:ymax, xmin:xmax].copy() if self._target_size is not None: image = misc.imresize(image, self._target_size) return image
def _read(self, idx): image = datasets.imread_rgb(os.path.join(self._root, 'images',\ self._raw_name[idx])) self._raw_dimension[idx] = image.shape[:2] xmin, ymin, xmax, ymax = \ self._get_cropped_coordinates(idx) image = image[ymin:ymax, xmin:xmax].copy() if self._target_size is not None: image = misc.imresize(image, self._target_size) return image
def read_image(self, name): """Reads the image and does the manipulation """ img = datasets.imread_rgb(name) return datasets.manipulate(img, None, None, None, CENTER_CROP)