def setUp(self):
self.tmp_dir = rv_config.get_tmp_dir()
self.class_config = ClassConfig(names=['one', 'two'])
self.class_config.update()
self.class_config.ensure_null_class()
self.null_class_id = self.class_config.get_null_class_id()
def _test_class_inf(self, props, exp_class_ids, default_class_id=None):
geojson = {
'type':
'FeatureCollection',
'features': [{
'properties': props,
'geometry': {
'type': 'Point',
'coordinates': [1, 1]
}
}]
}
json_to_file(geojson, self.uri)
class_config = ClassConfig(names=['building', 'car', 'tree'])
class_id_to_filter = {
0: ['==', 'type', 'building'],
1: ['any', ['==', 'type', 'car'], ['==', 'type', 'auto']]
}
vs_cfg = GeoJSONVectorSourceConfig(
uri=self.uri,
class_id_to_filter=class_id_to_filter,
default_class_id=default_class_id)
vs = vs_cfg.build(class_config, IdentityCRSTransformer())
trans_geojson = vs.get_geojson()
class_ids = [
f['properties']['class_id'] for f in trans_geojson['features']
]
self.assertEqual(class_ids, exp_class_ids)
def test_accounts_for_aoi(self):
class_config = ClassConfig(names=['car', 'building', 'background'])
label_source_uri = data_file_path('evaluator/cc-label-filtered.json')
label_source_cfg = ChipClassificationLabelSourceConfig(
vector_source=GeoJSONVectorSourceConfig(
uri=label_source_uri, default_class_id=None))
label_store_uri = data_file_path('evaluator/cc-label-full.json')
label_store_cfg = ChipClassificationGeoJSONStoreConfig(
uri=label_store_uri)
raster_source_uri = data_file_path('evaluator/cc-label-img-blank.tif')
raster_source_cfg = RasterioSourceConfig(uris=[raster_source_uri])
aoi_uri = data_file_path('evaluator/cc-label-aoi.json')
s = SceneConfig(
id='test',
raster_source=raster_source_cfg,
label_source=label_source_cfg,
label_store=label_store_cfg,
aoi_uris=[aoi_uri])
with rv_config.get_tmp_dir() as tmp_dir:
scene = s.build(class_config, tmp_dir)
output_uri = os.path.join(tmp_dir, 'eval.json')
evaluator = ChipClassificationEvaluatorConfig(
output_uri=output_uri).build(class_config)
evaluator.process([scene], tmp_dir)
overall = file_to_json(output_uri)['overall']
for item in overall:
self.assertEqual(item['f1'], 1.0)
def test_get_labels_rgb(self):
data = np.zeros((10, 10, 3), dtype=np.uint8)
data[7:, 7:, :] = [1, 1, 1]
null_class_id = 2
raster_source = MockRasterSource([0, 1, 2], 3)
raster_source.set_raster(data)
rgb_class_config = ClassConfig(names=['a'], colors=['#010101'])
rgb_class_config.ensure_null_class()
label_source = SemanticSegmentationLabelSource(
raster_source, null_class_id, rgb_class_config=rgb_class_config)
with label_source.activate():
window = Box.make_square(7, 7, 3)
labels = label_source.get_labels(window=window)
label_arr = labels.get_label_arr(window)
expected_label_arr = np.zeros((3, 3))
np.testing.assert_array_equal(label_arr, expected_label_arr)
def test_rgb_class_config_null_class(self):
raster_source_cfg = RasterioSourceConfig(uris=['/abc/def.tif'])
rgb_class_config = ClassConfig(names=['a'], colors=['#010101'])
cfg = SemanticSegmentationLabelSourceConfig(
raster_source=raster_source_cfg, rgb_class_config=rgb_class_config)
cfg.update()
self.assertEqual(len(cfg.rgb_class_config), 2)
self.assertIn('null', cfg.rgb_class_config.names)
def test_compute_ignore_class(self):
class_config = ClassConfig(names=['one', 'two'])
class_config.update()
class_config.ensure_null_class()
null_class_id = class_config.get_null_class_id()
gt_array = np.zeros((4, 4, 1), dtype=np.uint8)
gt_array[0, 0, 0] = 2
gt_raster = MockRasterSource([0], 1)
gt_raster.set_raster(gt_array)
gt_label_source = SemanticSegmentationLabelSource(
gt_raster, null_class_id)
pred_array = np.zeros((4, 4, 1), dtype=np.uint8)
pred_array[0, 0, 0] = 1
pred_raster = MockRasterSource([0], 1)
pred_raster.set_raster(pred_array)
pred_label_source = SemanticSegmentationLabelSource(
pred_raster, null_class_id)
eval = SemanticSegmentationEvaluation(class_config)
eval.compute(gt_label_source.get_labels(),
pred_label_source.get_labels())
self.assertAlmostEqual(1.0, eval.class_to_eval_item[0].f1)
self.assertAlmostEqual(1.0, eval.avg_item.f1)
def setUp(self):
self.crs_transformer = IdentityCRSTransformer()
self.extent = Box.make_square(0, 0, 10)
self.tmp_dir_obj = rv_config.get_tmp_dir()
self.tmp_dir = self.tmp_dir_obj.name
self.class_id = 0
self.background_class_id = 1
self.line_buffer = 1
self.class_config = ClassConfig(names=['a'])
self.uri = join(self.tmp_dir, 'tmp.json')
def setUp(self):
self.crs_transformer = DoubleCRSTransformer()
self.geojson = {
'type':
'FeatureCollection',
'features': [{
'type': 'Feature',
'geometry': {
'type':
'MultiPolygon',
'coordinates': [[[[0., 0.], [0., 2.], [2., 2.], [2., 0.],
[0., 0.]]]]
},
'properties': {
'class_name': 'car',
'class_id': 0,
'score': 0.0
}
}, {
'type': 'Feature',
'geometry': {
'type':
'Polygon',
'coordinates': [[[2., 2.], [2., 4.], [4., 4.], [4., 2.],
[2., 2.]]]
},
'properties': {
'score': 0.0,
'class_name': 'house',
'class_id': 1
}
}]
}
self.class_config = ClassConfig(names=['car', 'house'])
self.box1 = Box.make_square(0, 0, 4)
self.box2 = Box.make_square(4, 4, 4)
self.class_id1 = 0
self.class_id2 = 1
self.background_class_id = 2
geoms = []
for f in self.geojson['features']:
g = shape(f['geometry'])
g.class_id = f['properties']['class_id']
geoms.append(g)
self.str_tree = STRtree(geoms)
self.file_name = 'labels.json'
self.tmp_dir = rv_config.get_tmp_dir()
self.uri = os.path.join(self.tmp_dir.name, self.file_name)
json_to_file(self.geojson, self.uri)
def test_compute(self):
class_config = ClassConfig(names=['one', 'two'])
class_config.update()
class_config.ensure_null_class()
null_class_id = class_config.get_null_class_id()
gt_array = np.zeros((4, 4, 1), dtype=np.uint8)
gt_array[2, 2, 0] = 1
gt_array[0, 0, 0] = 2
gt_raster = MockRasterSource([0], 1)
gt_raster.set_raster(gt_array)
gt_label_source = SemanticSegmentationLabelSource(
gt_raster, null_class_id)
p_array = np.zeros((4, 4, 1), dtype=np.uint8)
p_array[1, 1, 0] = 1
p_raster = MockRasterSource([0], 1)
p_raster.set_raster(p_array)
p_label_source = SemanticSegmentationLabelSource(
p_raster, null_class_id)
eval = SemanticSegmentationEvaluation(class_config)
eval.compute(gt_label_source.get_labels(), p_label_source.get_labels())
tp0 = 16 - 3 # 4*4 - 3 true positives for class 0
fp0 = 1 # 1 false positive (2,2) and one don't care at (0,0)
fn0 = 1 # one false negative (1,1)
precision0 = float(tp0) / (tp0 + fp0)
recall0 = float(tp0) / (tp0 + fn0)
f10 = 2 * float(precision0 * recall0) / (precision0 + recall0)
tp1 = 0 # 0 true positives for class 1
fn1 = 1 # one false negative (2,2)
precision1 = 0 # float(tp1) / (tp1 + fp1) where fp1 == 1
recall1 = float(tp1) / (tp1 + fn1)
f11 = None
self.assertAlmostEqual(precision0,
eval.class_to_eval_item[0].precision)
self.assertAlmostEqual(recall0, eval.class_to_eval_item[0].recall)
self.assertAlmostEqual(f10, eval.class_to_eval_item[0].f1)
self.assertEqual(precision1, eval.class_to_eval_item[1].precision)
self.assertAlmostEqual(recall1, eval.class_to_eval_item[1].recall)
self.assertAlmostEqual(f11, eval.class_to_eval_item[1].f1)
avg_conf_mat = np.array([[0, 0, 0], [13., 1, 0], [1, 0, 0]])
avg_recall = (14 / 15) * recall0 + (1 / 15) * recall1
self.assertTrue(np.array_equal(avg_conf_mat, eval.avg_item.conf_mat))
self.assertEqual(avg_recall, eval.avg_item.recall)
def transform_geojson(self,
geojson,
line_bufs=None,
point_bufs=None,
crs_transformer=None,
to_map_coords=False):
if crs_transformer is None:
crs_transformer = IdentityCRSTransformer()
class_config = ClassConfig(names=['building'])
json_to_file(geojson, self.uri)
cfg = GeoJSONVectorSourceConfig(
uri=self.uri,
line_bufs=line_bufs,
point_bufs=point_bufs,
default_class_id=0)
source = cfg.build(class_config, crs_transformer)
return source.get_geojson(to_map_coords=to_map_coords)
def setUp(self):
self.file_name = 'labels.json'
self.tmp_dir = rv_config.get_tmp_dir()
self.file_path = os.path.join(self.tmp_dir.name, self.file_name)
self.crs_transformer = DoubleCRSTransformer()
self.geojson = {
'type':
'FeatureCollection',
'features': [{
'type': 'Feature',
'geometry': {
'type':
'Polygon',
'coordinates': [[[0., 0.], [0., 1.], [1., 1.], [1., 0.],
[0., 0.]]]
},
'properties': {
'class_id': 0,
'score': 0.9
}
}, {
'type': 'Feature',
'geometry': {
'type':
'Polygon',
'coordinates': [[[1., 1.], [1., 2.], [2., 2.], [2., 1.],
[1., 1.]]]
},
'properties': {
'score': 0.9,
'class_id': 1
}
}]
}
self.extent = Box.make_square(0, 0, 10)
self.class_config = ClassConfig(names=['car', 'house'])
json_to_file(self.geojson, self.file_path)
def test_vector_compute(self):
class_config = ClassConfig(names=['one', 'two'])
class_config.update()
class_config.ensure_null_class()
gt_uri = data_file_path('2-gt-polygons.geojson')
pred_uri = data_file_path('2-pred-polygons.geojson')
eval = SemanticSegmentationEvaluation(class_config)
eval.compute_vector(gt_uri, pred_uri, 'polygons', 0)
# NOTE: The two geojson files referenced above contain three
# unique geometries total, each file contains two geometries,
# and there is one geometry shared between the two.
tp = 1.0
fp = 1.0
fn = 1.0
precision = float(tp) / (tp + fp)
recall = float(tp) / (tp + fn)
self.assertAlmostEqual(precision, eval.class_to_eval_item[0].precision)
self.assertAlmostEqual(recall, eval.class_to_eval_item[0].recall)
def get_config(runner, root_uri, data_uri=None, full_train=False):
def get_path(part):
if full_train:
return join(data_uri, part)
else:
return join(dirname(__file__), part)
class_config = ClassConfig(names=['red', 'green'], colors=['red', 'green'])
def make_scene(id, img_path, label_path):
raster_source = RasterioSourceConfig(channel_order=[0, 1, 2],
uris=[img_path])
label_source = SemanticSegmentationLabelSourceConfig(
rgb_class_config=class_config,
raster_source=RasterioSourceConfig(uris=[label_path]))
label_store = SemanticSegmentationLabelStoreConfig(
rgb=True,
vector_output=[
PolygonVectorOutputConfig(class_id=0),
BuildingVectorOutputConfig(class_id=1)
])
return SceneConfig(id=id,
raster_source=raster_source,
label_source=label_source,
label_store=label_store)
chip_sz = 300
img_sz = chip_sz
data = SemanticSegmentationImageDataConfig(img_sz=img_sz, augmentors=[])
if full_train:
model = SemanticSegmentationModelConfig(backbone=Backbone.resnet50)
solver = SolverConfig(lr=1e-4,
num_epochs=300,
batch_sz=8,
one_cycle=True,
sync_interval=300)
else:
pretrained_uri = (
'https://github.com/azavea/raster-vision-data/releases/download/v0.12/'
'semantic-segmentation.pth')
model = SemanticSegmentationModelConfig(backbone=Backbone.resnet50,
init_weights=pretrained_uri)
solver = SolverConfig(lr=1e-9,
num_epochs=1,
batch_sz=2,
one_cycle=True,
sync_interval=200)
backend = PyTorchSemanticSegmentationConfig(data=data,
model=model,
solver=solver,
log_tensorboard=False,
run_tensorboard=False)
scenes = [
make_scene('test-scene', get_path('scene/image.tif'),
get_path('scene/labels.tif')),
make_scene('test-scene2', get_path('scene/image2.tif'),
get_path('scene/labels2.tif'))
]
scene_dataset = DatasetConfig(class_config=class_config,
train_scenes=scenes,
validation_scenes=scenes)
chip_options = SemanticSegmentationChipOptions(
window_method=SemanticSegmentationWindowMethod.sliding, stride=chip_sz)
return SemanticSegmentationConfig(root_uri=root_uri,
dataset=scene_dataset,
backend=backend,
train_chip_sz=chip_sz,
predict_chip_sz=chip_sz,
chip_options=chip_options)
def get_config(runner,
raw_uri: str,
processed_uri: str,
root_uri: str,
test: bool = False) -> SemanticSegmentationConfig:
"""Generate the pipeline config for this task. This function will be called
by RV, with arguments from the command line, when this example is run.
Args:
runner (Runner): Runner for the pipeline.
raw_uri (str): Directory where the raw data resides
processed_uri (str): Directory for storing processed data.
E.g. crops for testing.
root_uri (str): Directory where all the output will be written.
test (bool, optional): Whether to run in test mode. Defaults to False.
Returns:
SemanticSegmentationConfig: A pipeline config.
"""
if not test:
train_ids, val_ids = TRAIN_IDS, VAL_IDS
else:
train_ids, val_ids = TEST_MODE_TRAIN_IDS, TEST_MODE_VAL_IDS
raw_uri = UriPath(raw_uri)
processed_uri = UriPath(processed_uri)
# -------------------------------
# Configure dataset generation
# -------------------------------
class_config = ClassConfig(names=CLASS_NAMES, colors=CLASS_COLORS)
chip_options = SemanticSegmentationChipOptions(
window_method=SemanticSegmentationWindowMethod.sliding,
stride=CHIP_SIZE)
_make_scene = partial(make_scene,
raw_uri,
processed_uri,
class_config,
test_mode=test)
dataset_config = DatasetConfig(
class_config=class_config,
train_scenes=[_make_scene(scene_id) for scene_id in train_ids],
validation_scenes=[_make_scene(scene_id) for scene_id in val_ids])
# --------------------------------------------
# Configure PyTorch backend and training
# --------------------------------------------
model_config = SemanticSegmentationModelConfig(backbone=Backbone.resnet50)
solver_config = SolverConfig(lr=LR,
num_epochs=NUM_EPOCHS,
batch_sz=BATCH_SIZE,
test_num_epochs=TEST_MODE_NUM_EPOCHS,
test_batch_sz=TEST_MODE_BATCH_SIZE,
one_cycle=ONE_CYCLE)
backend_config = PyTorchSemanticSegmentationConfig(
model=model_config,
solver=solver_config,
log_tensorboard=LOG_TENSORBOARD,
run_tensorboard=RUN_TENSORBOARD,
test_mode=test)
# -----------------------------------------------
# Pass configurations to the pipeline config
# -----------------------------------------------
pipeline_config = SemanticSegmentationConfig(
root_uri=root_uri,
train_chip_sz=CHIP_SIZE,
predict_chip_sz=CHIP_SIZE,
chip_options=chip_options,
dataset=dataset_config,
backend=backend_config,
channel_display_groups=CHANNEL_DISPLAY_GROUPS)
return pipeline_config
def get_config(runner,
root_uri,
analyze_uri,
chip_uri,
json,
chip_sz=512,
batch_sz=32,
epochs=33,
preshrink=1,
small_test=False,
architecture='cheaplab',
level='L1C'):
chip_sz = int(chip_sz)
epochs = int(epochs)
batch_sz = int(batch_sz)
preshrink = int(preshrink)
assert (architecture in ['cheaplab', 'fpn-resnet18'])
assert (level in ['L1C', 'L2A'])
if level == 'L1C':
channel_order = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]
elif level == 'L2A':
channel_order = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]
num_channels = len(channel_order)
class_config = ClassConfig(names=["background", "cloud"],
colors=["brown", "white"])
class_id_filter_dict = {
0: ['==', 'default', 'Background'],
1: ['==', 'default', 'Cloud'],
}
train_crops = []
val_crops = []
for x in range(0, 5):
for y in range(0, 5):
x_start = x / 5.0
x_end = 0.80 - x_start
y_start = y / 5.0
y_end = 0.80 - y_start
crop = [x_start, y_start, x_end, y_end]
if x == y:
val_crops.append(crop)
else:
train_crops.append(crop)
scenes = get_scenes(json,
channel_order,
class_config,
class_id_filter_dict,
level,
train_crops=train_crops,
val_crops=val_crops)
train_scenes, validation_scenes = scenes
if small_test:
train_scenes = train_scenes[0:2]
validation_scenes = validation_scenes[0:2]
print(f"{len(train_scenes)} training scenes")
print(f"{len(validation_scenes)} validation scenes")
dataset = DatasetConfig(
class_config=class_config,
train_scenes=train_scenes,
validation_scenes=validation_scenes,
)
if architecture == 'fpn-resnet18':
external_def = ExternalModuleConfig(
github_repo='AdeelH/pytorch-fpn:0.1',
name='pytorch-fpn',
entrypoint='make_segm_fpn_resnet',
entrypoint_kwargs={
'name': 'resnet18',
'fpn_type': 'fpn',
'num_classes': 2,
'fpn_channels': 256,
'in_channels': len(channel_order),
'out_size': (chip_sz, chip_sz)
})
else:
external_def = ExternalModuleConfig(
github_repo='jamesmcclain/CheapLab:08d260b',
name='cheaplab',
entrypoint='make_cheaplab_model',
entrypoint_kwargs={
'preshrink': preshrink,
'num_channels': num_channels
})
model = SemanticSegmentationModelConfig(external_def=external_def)
external_loss_def = ExternalModuleConfig(
github_repo='jamesmcclain/CheapLab:08d260b',
name='bce_loss',
entrypoint='make_bce_loss',
force_reload=False,
entrypoint_kwargs={})
backend = PyTorchSemanticSegmentationConfig(
model=model,
solver=SolverConfig(lr=1e-4,
num_epochs=epochs,
batch_sz=batch_sz,
external_loss_def=external_loss_def,
ignore_last_class='force'),
log_tensorboard=False,
run_tensorboard=False,
num_workers=0,
preview_batch_limit=8)
chip_options = SemanticSegmentationChipOptions(
window_method=SemanticSegmentationWindowMethod.sliding, stride=chip_sz)
return SemanticSegmentationConfig(root_uri=root_uri,
analyze_uri=analyze_uri,
chip_uri=chip_uri,
dataset=dataset,
backend=backend,
train_chip_sz=chip_sz,
predict_chip_sz=chip_sz,
chip_options=chip_options,
chip_nodata_threshold=.75,
img_format='npy',
label_format='npy')
def get_config(runner, root_uri, data_uri=None, full_train=False):
def get_path(part):
if full_train:
return join(data_uri, part)
else:
return join(dirname(__file__), part)
class_config = ClassConfig(names=['car', 'building', 'background'],
colors=['red', 'blue', 'black'])
def make_scene(img_path, label_path):
id = basename(img_path)
label_source = ChipClassificationLabelSourceConfig(
vector_source=GeoJSONVectorSourceConfig(uri=label_path,
default_class_id=None,
ignore_crs_field=True),
ioa_thresh=0.5,
use_intersection_over_cell=False,
pick_min_class_id=True,
background_class_id=2,
infer_cells=True)
raster_source = RasterioSourceConfig(
channel_order=[0, 1, 2],
uris=[img_path],
transformers=[StatsTransformerConfig()])
return SceneConfig(id=id,
raster_source=raster_source,
label_source=label_source)
scenes = [
make_scene(get_path('scene/image.tif'), get_path('scene/labels.json')),
make_scene(get_path('scene/image2.tif'),
get_path('scene/labels2.json'))
]
scene_dataset = DatasetConfig(class_config=class_config,
train_scenes=scenes,
validation_scenes=scenes)
chip_sz = 200
img_sz = chip_sz
data = ClassificationImageDataConfig(img_sz=img_sz, augmentors=[])
if full_train:
model = ClassificationModelConfig(backbone=Backbone.resnet18)
solver = SolverConfig(lr=1e-4,
num_epochs=300,
batch_sz=8,
one_cycle=True,
sync_interval=300)
else:
pretrained_uri = (
'https://github.com/azavea/raster-vision-data/releases/download/v0.12/'
'chip-classification.pth')
model = ClassificationModelConfig(backbone=Backbone.resnet18,
init_weights=pretrained_uri)
solver = SolverConfig(lr=1e-9,
num_epochs=1,
batch_sz=2,
one_cycle=True,
sync_interval=200)
backend = PyTorchChipClassificationConfig(data=data,
model=model,
solver=solver,
log_tensorboard=False,
run_tensorboard=False)
config = ChipClassificationConfig(root_uri=root_uri,
dataset=scene_dataset,
backend=backend,
train_chip_sz=chip_sz,
predict_chip_sz=chip_sz)
return config
class TestSemanticSegmentationEvaluator(unittest.TestCase):
def setUp(self):
self.tmp_dir = rv_config.get_tmp_dir()
self.class_config = ClassConfig(names=['one', 'two'])
self.class_config.update()
self.class_config.ensure_null_class()
self.null_class_id = self.class_config.get_null_class_id()
def tearDown(self):
self.tmp_dir.cleanup()
def get_scene(self, class_id):
# Make scene where ground truth is all set to class_id
# and predictions are set to half 0's and half 1's
scene_id = str(class_id)
rs = MockRasterSource(channel_order=[0, 1, 2], num_channels=3)
rs.set_raster(np.zeros((10, 10, 3)))
gt_rs = MockRasterSource(channel_order=[0], num_channels=1)
gt_arr = np.full((10, 10, 1), class_id)
gt_rs.set_raster(gt_arr)
gt_ls = SemanticSegmentationLabelSource(gt_rs, self.null_class_id)
pred_rs = MockRasterSource(channel_order=[0], num_channels=1)
pred_arr = np.zeros((10, 10, 1))
pred_arr[5:10, :, :] = 1
pred_rs.set_raster(pred_arr)
pred_ls = SemanticSegmentationLabelSource(pred_rs, self.null_class_id)
return Scene(scene_id, rs, gt_ls, pred_ls)
def test_evaluator(self):
output_uri = join(self.tmp_dir.name, 'out.json')
scenes = [self.get_scene(0), self.get_scene(1)]
evaluator = SemanticSegmentationEvaluator(self.class_config,
output_uri, None)
evaluator.process(scenes, self.tmp_dir.name)
eval_json = file_to_json(output_uri)
exp_eval_json = file_to_json(data_file_path('expected-eval.json'))
self.assertDictEqual(eval_json, exp_eval_json)
def get_vector_scene(self, class_id, use_aoi=False):
gt_uri = data_file_path('{}-gt-polygons.geojson'.format(class_id))
pred_uri = data_file_path('{}-pred-polygons.geojson'.format(class_id))
scene_id = str(class_id)
rs = MockRasterSource(channel_order=[0, 1, 3], num_channels=3)
rs.set_raster(np.zeros((10, 10, 3)))
crs_transformer = IdentityCRSTransformer()
extent = Box.make_square(0, 0, 360)
config = RasterizedSourceConfig(
vector_source=GeoJSONVectorSourceConfig(uri=gt_uri,
default_class_id=0),
rasterizer_config=RasterizerConfig(background_class_id=1))
gt_rs = config.build(self.class_config, crs_transformer, extent)
gt_ls = SemanticSegmentationLabelSource(gt_rs, self.null_class_id)
config = RasterizedSourceConfig(
vector_source=GeoJSONVectorSourceConfig(uri=pred_uri,
default_class_id=0),
rasterizer_config=RasterizerConfig(background_class_id=1))
pred_rs = config.build(self.class_config, crs_transformer, extent)
pred_ls = SemanticSegmentationLabelSource(pred_rs, self.null_class_id)
pred_ls.vector_output = [
PolygonVectorOutputConfig(uri=pred_uri,
denoise=0,
class_id=class_id)
]
if use_aoi:
aoi_uri = data_file_path('{}-aoi.geojson'.format(class_id))
aoi_geojson = file_to_json(aoi_uri)
aoi_polygons = [shape(aoi_geojson['features'][0]['geometry'])]
return Scene(scene_id, rs, gt_ls, pred_ls, aoi_polygons)
return Scene(scene_id, rs, gt_ls, pred_ls)
def test_vector_evaluator(self):
output_uri = join(self.tmp_dir.name, 'raster-out.json')
vector_output_uri = join(self.tmp_dir.name, 'vector-out.json')
scenes = [self.get_vector_scene(0), self.get_vector_scene(1)]
evaluator = SemanticSegmentationEvaluator(self.class_config,
output_uri,
vector_output_uri)
evaluator.process(scenes, self.tmp_dir.name)
vector_eval_json = file_to_json(vector_output_uri)
exp_vector_eval_json = file_to_json(
data_file_path('expected-vector-eval.json'))
# NOTE: The precision and recall values found in the file
# `expected-vector-eval.json` are equal to fractions of the
# form (n-1)/n for n <= 7 which can be seen to be (and have
# been manually verified to be) correct.
self.assertDictEqual(vector_eval_json, exp_vector_eval_json)
def test_vector_evaluator_with_aoi(self):
output_uri = join(self.tmp_dir.name, 'raster-out.json')
vector_output_uri = join(self.tmp_dir.name, 'vector-out.json')
scenes = [self.get_vector_scene(0, use_aoi=True)]
evaluator = SemanticSegmentationEvaluator(self.class_config,
output_uri,
vector_output_uri)
evaluator.process(scenes, self.tmp_dir.name)
vector_eval_json = file_to_json(vector_output_uri)
exp_vector_eval_json = file_to_json(
data_file_path('expected-vector-eval-with-aoi.json'))
# NOTE: The precision and recall values found in the file
# `expected-vector-eval.json` are equal to fractions of the
# form (n-1)/n for n <= 7 which can be seen to be (and have
# been manually verified to be) correct.
self.assertDictEqual(vector_eval_json, exp_vector_eval_json)
def get_config(runner, root_uri, data_uri=None, full_train=False):
def get_path(part):
if full_train:
return join(data_uri, part)
else:
return join(dirname(__file__), part)
class_config = ClassConfig(names=['car', 'building'],
colors=['blue', 'red'])
def make_scene(scene_id, img_path, label_path):
raster_source = RasterioSourceConfig(channel_order=[0, 1, 2],
uris=[img_path])
label_source = ObjectDetectionLabelSourceConfig(
vector_source=GeoJSONVectorSourceConfig(uri=label_path,
default_class_id=None))
return SceneConfig(id=scene_id,
raster_source=raster_source,
label_source=label_source)
chip_sz = 300
img_sz = chip_sz
data = ObjectDetectionImageDataConfig(img_sz=img_sz, augmentors=[])
if full_train:
model = ObjectDetectionModelConfig(backbone=Backbone.resnet18)
solver = SolverConfig(lr=1e-4,
num_epochs=300,
batch_sz=8,
one_cycle=True,
sync_interval=300)
else:
pretrained_uri = (
'https://github.com/azavea/raster-vision-data/releases/download/v0.12/'
'object-detection.pth')
model = ObjectDetectionModelConfig(backbone=Backbone.resnet18,
init_weights=pretrained_uri)
solver = SolverConfig(lr=1e-9,
num_epochs=1,
batch_sz=2,
one_cycle=True,
sync_interval=200)
backend = PyTorchObjectDetectionConfig(data=data,
model=model,
solver=solver,
log_tensorboard=False,
run_tensorboard=False)
scenes = [
make_scene('od_test', get_path('scene/image.tif'),
get_path('scene/labels.json')),
make_scene('od_test-2', get_path('scene/image2.tif'),
get_path('scene/labels2.json'))
]
scene_dataset = DatasetConfig(class_config=class_config,
train_scenes=scenes,
validation_scenes=scenes)
chip_options = ObjectDetectionChipOptions(neg_ratio=1.0, ioa_thresh=1.0)
predict_options = ObjectDetectionPredictOptions(merge_thresh=0.1,
score_thresh=0.5)
return ObjectDetectionConfig(root_uri=root_uri,
dataset=scene_dataset,
backend=backend,
train_chip_sz=chip_sz,
predict_chip_sz=chip_sz,
chip_options=chip_options,
predict_options=predict_options)
def get_config(runner,
raw_uri: str,
processed_uri: str,
root_uri: str,
nochip: bool = True,
test: bool = False) -> SemanticSegmentationConfig:
"""Generate the pipeline config for this task. This function will be called
by RV, with arguments from the command line, when this example is run.
Args:
runner (Runner): Runner for the pipeline.
raw_uri (str): Directory where the raw data resides
processed_uri (str): Directory for storing processed data.
E.g. crops for testing.
root_uri (str): Directory where all the output will be written.
nochip (bool, optional): If True, read directly from the TIFF during
training instead of from pre-generated chips. The analyze and chip
commands should not be run, if this is set to True. Defaults to
True.
test (bool, optional): If True, does the following simplifications:
(1) Uses only the first 2 scenes
(2) Uses only a 600x600 crop of the scenes
(3) Enables test mode in the learner, which makes it use the
test_batch_sz and test_num_epochs, among other things.
Defaults to False.
Returns:
SemanticSegmentationConfig: A pipeline config.
"""
if not test:
train_ids, val_ids = TRAIN_IDS, VAL_IDS
else:
train_ids, val_ids = TEST_MODE_TRAIN_IDS, TEST_MODE_VAL_IDS
raw_uri = UriPath(raw_uri)
processed_uri = UriPath(processed_uri)
# -------------------------------
# Configure dataset generation
# -------------------------------
class_config = ClassConfig(names=CLASS_NAMES, colors=CLASS_COLORS)
_make_scene = partial(make_scene,
raw_uri,
processed_uri,
class_config,
test_mode=test)
dataset_config = DatasetConfig(
class_config=class_config,
train_scenes=[_make_scene(scene_id) for scene_id in train_ids],
validation_scenes=[_make_scene(scene_id) for scene_id in val_ids],
img_channels=5)
chip_options = SemanticSegmentationChipOptions(
window_method=SemanticSegmentationWindowMethod.sliding,
stride=CHIP_SIZE)
if nochip:
window_opts = GeoDataWindowConfig(method=GeoDataWindowMethod.sliding,
size=CHIP_SIZE,
stride=chip_options.stride)
data = SemanticSegmentationGeoDataConfig(
scene_dataset=dataset_config,
window_opts=window_opts,
img_sz=CHIP_SIZE,
num_workers=4,
channel_display_groups=CHANNEL_DISPLAY_GROUPS)
else:
data = SemanticSegmentationImageDataConfig(
img_sz=CHIP_SIZE,
num_workers=4,
channel_display_groups=CHANNEL_DISPLAY_GROUPS)
# --------------------------------------------
# Configure PyTorch backend and training
# --------------------------------------------
model_config = SemanticSegmentationModelConfig(backbone=Backbone.resnet50)
solver_config = SolverConfig(lr=LR,
num_epochs=NUM_EPOCHS,
batch_sz=BATCH_SIZE,
test_num_epochs=TEST_MODE_NUM_EPOCHS,
test_batch_sz=TEST_MODE_BATCH_SIZE,
one_cycle=ONE_CYCLE)
backend_config = PyTorchSemanticSegmentationConfig(
data=data,
model=model_config,
solver=solver_config,
log_tensorboard=LOG_TENSORBOARD,
run_tensorboard=RUN_TENSORBOARD,
test_mode=test)
# -----------------------------------------------
# Pass configurations to the pipeline config
# -----------------------------------------------
pipeline_config = SemanticSegmentationConfig(
root_uri=root_uri,
train_chip_sz=CHIP_SIZE,
predict_chip_sz=CHIP_SIZE,
chip_options=chip_options,
dataset=dataset_config,
backend=backend_config,
channel_display_groups=CHANNEL_DISPLAY_GROUPS)
return pipeline_config
def get_config(runner,
root_uri,
json,
dataset,
catalog_dir='/vsizip//workdir',
imagery_dir='/opt/data',
chip_sz=512,
N=None):
chip_sz = int(chip_sz)
if dataset == 'cloud':
class_config = ClassConfig(
names=[
'algal_bloom', 'normal_water', 'cloud', 'cloud_shadow', 'other'
],
colors=['green', 'blue', 'white', 'gray', 'brown'])
class_id_filter_dict = {
0: ['==', 'default', 'Algal bloom'],
1: ['==', 'default', 'Non-algal-bloomed water'],
2: ['==', 'default', 'Cloud'],
3: ['==', 'default', 'Cloud shadow'],
0xff: ['==', 'default', 'Other '],
}
elif dataset == 'tree':
class_config = ClassConfig(names=['green_stage', 'red_stage', 'other'],
colors=['green', 'brown', 'cyan'])
class_id_filter_dict = {
0: ['==', 'default', "Green stage conifer"],
1: ['==', 'default', "Red stage conifer"],
0xff: ['==', 'default', "Other"],
}
else:
raise Exception()
train_crops = []
val_crops = []
for x in range(0, 5):
for y in range(0, 5):
x_start = x / 5.0
x_end = 0.80 - x_start
y_start = y / 5.0
y_end = 0.80 - y_start
crop = [x_start, y_start, x_end, y_end]
if x == y:
val_crops.append(crop)
else:
train_crops.append(crop)
scenes = get_scenes(json,
class_config,
class_id_filter_dict,
catalog_dir,
imagery_dir,
train_crops=train_crops,
val_crops=val_crops,
N=N)
train_scenes, validation_scenes = scenes
print(f'{len(train_scenes)} training scenes')
print(f'{len(validation_scenes)} validation scenes')
dataset = DatasetConfig(
class_config=class_config,
train_scenes=train_scenes,
validation_scenes=validation_scenes,
)
solver = SolverConfig()
data = SemanticSegmentationImageDataConfig(img_sz=chip_sz,
num_workers=0,
preview_batch_limit=8)
model = SemanticSegmentationModelConfig()
backend = PyTorchSemanticSegmentationConfig(model=model,
data=data,
solver=solver)
chip_options = SemanticSegmentationChipOptions(
window_method=SemanticSegmentationWindowMethod.sliding, stride=chip_sz)
return SemanticSegmentationConfig(root_uri=root_uri,
dataset=dataset,
backend=backend,
train_chip_sz=chip_sz,
predict_chip_sz=chip_sz,
chip_options=chip_options,
chip_nodata_threshold=.75,
img_format='npy',
label_format='png')