def test_filter_by_aoi(self):
windows = [Box.make_square(0, 0, 2), Box.make_square(0, 2, 2)]
aoi_polygons = [Box.make_square(0, 0, 3).to_shapely()]
filt_windows = Box.filter_by_aoi(windows, aoi_polygons, within=False)
self.assertListEqual(filt_windows, windows)
filt_windows = Box.filter_by_aoi(windows, aoi_polygons, within=True)
self.assertListEqual(filt_windows, windows[0:1])
def init_windows(self) -> None:
"""Pre-compute windows."""
windows = self.scene.raster_source.get_extent().get_windows(
chip_sz=self.size, stride=self.stride, padding=self.padding)
if len(self.scene.aoi_polygons) > 0:
windows = Box.filter_by_aoi(windows, self.scene.aoi_polygons)
self.windows = windows
def filter_windows(windows: Sequence[Box]) -> List[Box]:
"""Filter out chips that
(1) are outside the AOI
(2) only consist of null labels
(3) have NODATA proportion >= chip_nodata_threshold
"""
total_windows = len(windows)
if scene.aoi_polygons:
windows = Box.filter_by_aoi(windows, scene.aoi_polygons)
log.info(f'AOI filtering: {len(windows)}/{total_windows} '
'chips accepted')
filt_windows = []
for w in windows:
chip = raster_source.get_chip(w)
nodata_below_thresh = nodata_below_threshold(
chip, chip_nodata_threshold, nodata_val=0)
label_arr = label_source.get_labels(w).get_label_arr(w)
null_labels = label_arr == class_config.get_null_class_id()
if not np.all(null_labels) and nodata_below_thresh:
filt_windows.append(w)
log.info('Label and NODATA filtering: '
f'{len(filt_windows)}/{len(windows)} chips accepted')
windows = filt_windows
return windows
def get_train_windows(scene, chip_size):
train_windows = []
extent = scene.raster_source.get_extent()
stride = chip_size
windows = extent.get_windows(chip_size, stride)
if scene.aoi_polygons:
windows = Box.filter_by_aoi(windows, scene.aoi_polygons)
for window in windows:
chip = scene.raster_source.get_chip(window)
if np.sum(chip.ravel()) > 0:
train_windows.append(window)
return train_windows
def filter_windows(windows):
if scene.aoi_polygons:
windows = Box.filter_by_aoi(windows, scene.aoi_polygons)
filt_windows = []
for w in windows:
label_arr = label_source.get_labels(w).get_label_arr(w)
null_inds = (
label_arr.ravel() == class_config.get_null_class_id())
if not np.all(null_inds):
filt_windows.append(w)
windows = filt_windows
return windows
def filter_windows(windows):
if scene.aoi_polygons:
windows = Box.filter_by_aoi(windows, scene.aoi_polygons)
if 0 in self.config.class_map.get_keys():
filt_windows = []
for w in windows:
label_arr = label_source.get_labels(w).get_label_arr(w)
ignore_inds = label_arr.ravel() == 0
if np.all(ignore_inds):
pass
else:
filt_windows.append(w)
windows = filt_windows
return windows
def get_train_windows(scene: Scene,
chip_size: int,
chip_nodata_threshold: float = 1.) -> List[Box]:
train_windows = []
extent = scene.raster_source.get_extent()
stride = chip_size
windows = extent.get_windows(chip_size, stride)
total_windows = len(windows)
if scene.aoi_polygons:
windows = Box.filter_by_aoi(windows, scene.aoi_polygons)
log.info(f'AOI filtering: {len(windows)}/{total_windows} '
'chips accepted')
for window in windows:
chip = scene.raster_source.get_chip(window)
if nodata_below_threshold(chip, chip_nodata_threshold, nodata_val=0):
train_windows.append(window)
log.info('NODATA filtering: '
f'{len(train_windows)}/{len(windows)} chips accepted')
return train_windows
def filter_windows(windows):
if scene.aoi_polygons:
windows = Box.filter_by_aoi(windows, scene.aoi_polygons)
return windows
