21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232 | class AggregatorComponent(BaseComponent):
"""
Aggregates overlapping detections/segmentations from tiled inference.
Requirements:
- infer_gdf with geometry, object_id, tile_path columns
- Score columns based on config weights (detector_score/segmenter_score)
Produces:
- Merged GeoDataFrame with aggregator_score
- GeoPackage files (aggregated + pre-aggregated)
- COCO file
"""
name = 'aggregator'
BASE_REQUIRES_STATE = {StateKey.INFER_GDF, StateKey.PRODUCT_NAME}
BASE_REQUIRES_COLUMNS = {Col.GEOMETRY, Col.OBJECT_ID, Col.TILE_PATH}
BASE_PRODUCES_STATE = {StateKey.INFER_GDF, StateKey.INFER_COCO_PATH}
BASE_PRODUCES_COLUMNS = {Col.AGGREGATOR_SCORE}
BASE_STATE_HINTS = {
StateKey.INFER_GDF: (
"Aggregator needs a GeoDataFrame with detections/segmentations. "
"Add a detector or segmenter before aggregator in the pipeline."
),
}
BASE_COLUMN_HINTS = {
Col.GEOMETRY: "GeoDataFrame must have a 'geometry' column with polygon geometries.",
Col.OBJECT_ID: "Each detection needs a unique 'canopyrs_object_id'. Created by detector/segmenter.",
Col.TILE_PATH: "Each detection needs a 'tile_path' column indicating source tile.",
}
def __init__(
self,
config: AggregatorConfig,
parent_output_path: str = None,
component_id: int = None
):
super().__init__(config, parent_output_path, component_id)
# Set base requirements
self.requires_state = set(self.BASE_REQUIRES_STATE)
self.requires_columns = set(self.BASE_REQUIRES_COLUMNS)
self.produces_state = set(self.BASE_PRODUCES_STATE)
self.produces_columns = set(self.BASE_PRODUCES_COLUMNS)
# Set hints
self.state_hints = dict(self.BASE_STATE_HINTS)
self.column_hints = dict(self.BASE_COLUMN_HINTS)
# Add config-dependent requirements
if config.detector_score_weight > 0:
self.requires_columns.add(Col.DETECTOR_SCORE)
self.column_hints[Col.DETECTOR_SCORE] = (
f"Config has detector_score_weight={config.detector_score_weight} > 0, "
f"so '{Col.DETECTOR_SCORE}' column is required. "
f"Add a detector before aggregator, or set detector_score_weight=0."
)
if config.segmenter_score_weight > 0:
self.requires_columns.add(Col.SEGMENTER_SCORE)
self.column_hints[Col.SEGMENTER_SCORE] = (
f"Config has segmenter_score_weight={config.segmenter_score_weight} > 0, "
f"so '{Col.SEGMENTER_SCORE}' column is required. "
f"Add a segmenter before aggregator, or set segmenter_score_weight=0."
)
@classmethod
def run_standalone(
cls,
config: AggregatorConfig,
infer_gdf: 'gpd.GeoDataFrame',
output_path: str,
product_name: str = "standalone",
) -> 'DataState':
"""
Run aggregator standalone on a GeoDataFrame of detections/segmentations.
Args:
config: Aggregator configuration
infer_gdf: GeoDataFrame with geometry, object_id, and tile_path columns
output_path: Where to save outputs
product_name: Name for output files (used in gpkg naming)
Returns:
DataState with aggregated results (access .infer_gdf for the GeoDataFrame)
Example:
result = AggregatorComponent.run_standalone(
config=AggregatorConfig(nms_threshold=0.5, ...),
infer_gdf=my_detections_gdf,
output_path='./output',
)
print(result.infer_gdf)
"""
from canopyrs.engine.pipeline import run_component
return run_component(
component=cls(config),
output_path=output_path,
infer_gdf=infer_gdf,
product_name=product_name,
)
@validate_requirements
def __call__(self, data_state: DataState) -> ComponentResult:
"""
Aggregate overlapping detections/segmentations.
Returns ComponentResult - Pipeline handles I/O and state updates.
"""
# Suppress geographic CRS area warnings from geopandas
warnings.filterwarnings('ignore', message='.*Geometry is in a geographic CRS.*')
infer_gdf = data_state.infer_gdf
columns_to_pass = data_state.infer_gdf_columns_to_pass
# Build score columns and weights from config
score_cols = []
weights = []
if self.config.detector_score_weight > 0:
score_cols.append(Col.DETECTOR_SCORE)
weights.append(self.config.detector_score_weight)
if self.config.segmenter_score_weight > 0:
score_cols.append(Col.SEGMENTER_SCORE)
weights.append(self.config.segmenter_score_weight)
# Generate output names
gpkg_name, pre_agg_gpkg_name = self._get_gpkg_names(data_state)
# Drop some previous components columns that can interfere with aggregation
for col in [Col.AGGREGATOR_SCORE, 'tile_id']:
if col in infer_gdf.columns:
infer_gdf = infer_gdf.drop(columns=[col])
if col in columns_to_pass:
columns_to_pass.remove(col)
# Run aggregation (geodataset Aggregator handles its own file saving)
aggregator = Aggregator.from_gdf(
output_path=self.output_path / gpkg_name if self.output_path else None,
gdf=infer_gdf,
tiles_paths_column=Col.TILE_PATH,
polygons_column=Col.GEOMETRY,
scores_column=score_cols if score_cols else None,
other_attributes_columns=list(columns_to_pass),
scores_weights=weights if weights else None,
scores_weighting_method=self.config.scores_weighting_method,
min_centroid_distance_weight=self.config.min_centroid_distance_weight,
score_threshold=self.config.score_threshold,
nms_threshold=self.config.nms_threshold,
nms_algorithm=self.config.nms_algorithm,
best_geom_keep_area_ratio=self.config.best_geom_keep_area_ratio,
edge_band_buffer_percentage=self.config.edge_band_buffer_percentage,
pre_aggregated_output_path=self.output_path / pre_agg_gpkg_name if self.output_path else None,
)
result_gdf = aggregator.polygons_gdf
# Determine category column for COCO
coco_categories_col = None
if Col.SEGMENTER_CLASS in result_gdf.columns:
coco_categories_col = Col.SEGMENTER_CLASS
elif Col.DETECTOR_CLASS in result_gdf.columns:
coco_categories_col = Col.DETECTOR_CLASS
# Register the GeoPackages that geodataset already wrote
output_files = {}
if self.output_path:
output_files['gpkg'] = self.output_path / gpkg_name
output_files['pre_aggregated_gpkg'] = self.output_path / pre_agg_gpkg_name
return ComponentResult(
gdf=result_gdf,
produced_columns=columns_to_pass | {Col.AGGREGATOR_SCORE},
objects_are_new=False,
save_gpkg=False, # Aggregator already saves via geodataset
save_coco=True,
coco_scores_column=Col.AGGREGATOR_SCORE,
coco_categories_column=coco_categories_col,
output_files=output_files,
)
def _get_gpkg_names(self, data_state: DataState) -> tuple:
"""Generate GeoPackage names using the product name from data state."""
try:
_, scale_factor, ground_resolution, _, _, _ = TileNameConvention().parse_name(
Path(data_state.infer_gdf[Col.TILE_PATH].iloc[0]).name
)
except Exception as e:
scale_factor = 1.0
ground_resolution = None
gpkg_name = GeoPackageNameConvention.create_name(
product_name=data_state.product_name,
fold=INFER_AOI_NAME,
scale_factor=scale_factor,
ground_resolution=ground_resolution
)
pre_agg_gpkg_name = GeoPackageNameConvention.create_name(
product_name=data_state.product_name,
fold=f'{INFER_AOI_NAME}notaggregated',
scale_factor=scale_factor,
ground_resolution=ground_resolution
)
return gpkg_name, pre_agg_gpkg_name
|