def main():
"""
this script extracts patches using images and lesion masks
images with the same name will be matched between the two input directories
CLI Args:
1: a directory of segmented lung images
2: a directory of lesion masks
3: the output directory for the patches
"""
image_dir = sys.argv[1]
mask_dir = sys.argv[2]
out_dir = sys.argv[3]
image_files = get_all_files(image_dir)
mask_files = get_all_files(mask_dir)
check_files = list(
set([ops(f)[0]
for f in image_files]) & set([ops(f)[0] for f in mask_files]))
for cf in check_files:
mask_file = [f for f in mask_files if cf == ops(f)[0]][0]
image_file = [f for f in image_files if cf == ops(f)[0]][0]
mask = Image.open(opj(mask_dir, mask_file))
#mask = PIL.ImageOps.invert(mask)
patch_coords = util.find_lesion_coordiates(np.array(mask))
image = Image.open(opj(image_dir, image_file))
#image = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2BGR)
#image = preprocess.preprocess(image)
#print(cf, patch_coords, image.shape)
patch = util.extract_patch(np.array(image), patch_coords,
preprocess.PATCH_SIZE)
patch_image = Image.fromarray(np.uint8(patch)).convert("L")
patch_image.save(opj(out_dir, cf + ".png"), "PNG")
def __call__(self, data):
count = 0
for fpath in _find_files(self.regex, dirs=self.dirs, topdir=self.topdir):
lgr.log(5, "Found file %s" % fpath)
count += 1
path, filename = ops(fpath)
yield updated(data, {'path': path, 'filename': filename})
self._total_count += count
if not self._total_count and self.fail_if_none:
raise RuntimeError("We did not match any file using regex %r" % self.regex)
def __call__(self, data):
count = 0
for fpath in _find_files(self.regex,
dirs=self.dirs,
topdir=self.topdir):
lgr.log(5, "Found file %s" % fpath)
count += 1
path, filename = ops(fpath)
yield updated(data, {'path': path, 'filename': filename})
self._total_count += count
if not self._total_count and self.fail_if_none:
raise RuntimeError("We did not match any file using regex %r" %
self.regex)
def commit_collection(self, collection, msg):
if self.is_read_only:
raise RuntimeWarning("Can't commit remote collection.")
if not isinstance(collection, Collection):
raise TypeError("Can't save non-collection type: %s" %
type(collection))
# save current branch ...
current_branch = self.repo.git_get_active_branch()
if self.branch != current_branch:
# ... and switch to the one to be changed:
self.repo.git_checkout(self.branch)
# handle files we no longer have:
files_to_remove = [f for f in self.repo.get_indexed_files()
if self.repo._filename2key(ops(f)[0]) not in
collection.keys()]
self.repo.git_remove(files_to_remove)
# update everything else to be safe
files_to_add = []
# collection level:
collection.meta.serialize(opj(self.repo.path, REPO_STD_META_FILE),
format="turtle")
files_to_add.append(REPO_STD_META_FILE)
# handles:
for k, v in collection.iteritems():
v.commit()
# files_to_add.append(self.repo._key2filename(k))
# Actually, this shouldn't be necessary, since it was
# committed above. On the other hand, that's a lot of commits.
# May be don't commit the handles but just write_to_file and commit
# herein.
self.repo.git_add(files_to_add)
self.repo.git_commit(msg)
if self.branch != current_branch:
# switch back to repo's active branch on disk
self.repo.git_checkout(current_branch)