swiss-territorial-data-lab · GwenaelleSa · Sep 25, 2025 · Aug 26, 2025 · Aug 28, 2025 · Aug 28, 2025
diff --git a/README.md b/README.md
@@ -232,8 +232,11 @@ train_model.py:
     tst: <the COCO JSON file related to the test dataset>
   detectron2_config_file: <the detectron2 configuration file (relative path w/ respect to the working_folder>
   model_weights:
-    model_zoo_checkpoint_url: <e.g. "COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml">
+    pth_file: <path to the model if training is resumed. Defaults to "<log directory>/model_final.pth">
+    model_zoo_checkpoint_url: <zoo model to start training from, e.g. "COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml">
     init_model_weights: <True or False; if True, the model weights will be initialized to 0 (optional, defaults to False)>
+  resume_training: <True or False; if True, the training is resumed from the final weights saved in the log folder. Defaults to False>
+  data_augmentation: <True or False; if True, apply random adjustment of brightness, contrast, saturation, lightning, and size, plus flip the image horizontally. Defaults to False>
 ```
 
 Detectron2's configuration files are provided in the example folders mentioned here-below. We warn the end-user about the fact that, **for the time being, no hyperparameters tuning is automatically performed**.
@@ -324,7 +327,7 @@ A few examples are provided within the `examples` folder. For further details, w
 * [Segmentation of Border Points based on Analog Cadastral Plans](examples/borderpoints/README.md): multi-class instance segmentation with images from another folder based on a custom grid,
 * [Evolution of Mineral Extraction Sites over the Entire Switzerland](examples/mineral-extrac-sites-detection/README.md): object monitoring with images from an XYZ service,
 * [Swimming Pool Detection over the Canton of Geneva](examples/swimming-pool-detection/GE/README.md): instance segmentation with images from a MIL service,
-* [Swimming Pool Detection over the Canton of Neuchâtel](examples/swimming-pool-detection/NE/README.md): instance segmentation with images from a WMS service.service,
+* [Swimming Pool Detection over the Canton of Neuchâtel](examples/swimming-pool-detection/NE/README.md): instance segmentation with images from a WMS service.
 
 It is brought to the reader attention that the examples are provided with a debug parameter that can be set to `True` for quick tests.
 

diff --git a/examples/anthropogenic-activities/README.md b/examples/anthropogenic-activities/README.md
@@ -1,4 +1,4 @@
-# Example: detection of anthropic soils potentially suitable for rehabilitation
+# Example: segmentation of anthropic soils on historic images
 
 A working setup is provided here to test the multi-class classification and the use of empty tiles, as well as false positive ones.
 It consists of the following elements:

diff --git a/examples/anthropogenic-activities/config_trne.yaml b/examples/anthropogenic-activities/config_trne.yaml
@@ -94,21 +94,10 @@ assess_detections.py:
   metrics_method: micro-average   # 1: macro-average ; 2: macro-weighted-average ; 3: micro-average
   # confidence_threshold: 0.05
 
-# Plots (optional)
-result_analysis.py:
-  working_directory: output/trne
-  output_directory: plots
-  detections: tagged_detections.gpkg
-  min_year: 1950
-  max_year: 2023
-  class_dict: {'Activité non agricole': 'Non-agricultural activity', # Provide a customed legend
-                'Mouvement de terrain': 'Land movement'}
-
 # Merge detections across tiles
 merge_detections.py:
   working_directory: output/trne
   output_dir: post_processed
-  labels: labels.geojson
   detections:
     trn: trn_detections_at_0dot05_threshold.gpkg
     val: val_detections_at_0dot05_threshold.gpkg

diff --git a/examples/anthropogenic-activities/detectron2_config.yaml b/examples/anthropogenic-activities/detectron2_config.yaml
@@ -262,7 +262,6 @@ MODEL:
     NAME: SemSegFPNHead
     NORM: GN
     NUM_CLASSES: 54
-  WEIGHTS: https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x/137260431/model_final_a54504.pkl
 OUTPUT_DIR: logs
 SEED: 42
 SOLVER:

diff --git a/examples/anthropogenic-activities/merge_detections.py b/examples/anthropogenic-activities/merge_detections.py
@@ -35,7 +35,6 @@
     # Load input parameters
     WORKING_DIR = cfg['working_directory']
     OUTPUT_DIR = cfg['output_dir']
-    LABELS = cfg['labels'] if 'labels' in cfg.keys() else None
     DETECTION_FILES = cfg['detections']
 
     DISTANCE = cfg['distance']
@@ -100,8 +99,8 @@
         detections_tiles_join_gdf = gpd.sjoin(tiles_gdf, detections_buffer_gdf, how='left', predicate='contains')
         remove_det_list = detections_tiles_join_gdf.det_id.unique().tolist()
 
-        detections_overlap_tiles_gdf = detections_by_year_gdf[~detections_by_year_gdf.det_id.isin(remove_det_list)].drop_duplicates(subset=['det_id'], ignore_index=True)
         detections_within_tiles_gdf = detections_by_year_gdf[detections_by_year_gdf.det_id.isin(remove_det_list)].drop_duplicates(subset=['det_id'], ignore_index=True)
+        detections_overlap_tiles_gdf = detections_by_year_gdf[~detections_by_year_gdf.det_id.isin(remove_det_list)].drop_duplicates(subset=['det_id'], ignore_index=True)
 
         # Merge polygons within the thd distance
         detections_overlap_tiles_gdf.loc[:, 'geometry'] = detections_overlap_tiles_gdf.buffer(DISTANCE, resolution=2)

diff --git a/examples/anthropogenic-activities/prepare_data.py b/examples/anthropogenic-activities/prepare_data.py
@@ -61,7 +61,7 @@
 
     written_files = []
 
-    gt_labels_4326_gdf = ffe.preapre_labels(SHPFILE, CATEGORY, supercategory=SUPERCATEGORY)
+    gt_labels_4326_gdf = ffe.prepare_labels(SHPFILE, CATEGORY, supercategory=SUPERCATEGORY)
 
     label_filepath = os.path.join(OUTPUT_DIR, 'labels.geojson')
     gt_labels_4326_gdf.to_file(label_filepath, driver='GeoJSON')

diff --git a/examples/borderpoints/config/detectron2_config.yaml b/examples/borderpoints/config/detectron2_config.yaml
@@ -258,8 +258,6 @@ MODEL:
     LOSS_WEIGHT: 1.0
     NAME: SemSegFPNHead
     NORM: GN
-    NUM_CLASSES: 54
-  WEIGHTS: https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x/137260431/model_final_a54504.pkl
 OUTPUT_DIR: logs
 SEED: 42
 SOLVER:

diff --git a/examples/mineral-extract-sites-detection/config_trne.yaml b/examples/mineral-extract-sites-detection/config_trne.yaml
@@ -57,7 +57,7 @@ train_model.py:
   detectron2_config_file: ../../detectron2_config_dqry.yaml # path relative to the working_folder
   model_weights:
     model_zoo_checkpoint_url: COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml
-    
+
 # Object detection with the optimised trained model
 make_detections.py:
   working_directory: ./output/trne/

diff --git a/examples/mineral-extract-sites-detection/detectron2_config_dqry.yaml b/examples/mineral-extract-sites-detection/detectron2_config_dqry.yaml
@@ -261,8 +261,6 @@ MODEL:
     LOSS_WEIGHT: 1.0
     NAME: SemSegFPNHead
     NORM: GN
-    NUM_CLASSES: 54
-  WEIGHTS: https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x/137260431/model_final_a54504.pkl
 OUTPUT_DIR: logs
 SEED: 42
 SOLVER:

diff --git a/examples/mineral-extract-sites-detection/prepare_data.py b/examples/mineral-extract-sites-detection/prepare_data.py
@@ -10,7 +10,7 @@
 import geopandas as gpd
 
 sys.path.insert(0, '../..')
-from helpers.functions_for_examples import format_all_tiles, preapre_labels
+from helpers.functions_for_examples import format_all_tiles, prepare_labels
 from helpers.misc import format_logger
 from helpers.constants import DONE_MSG
 
@@ -60,7 +60,7 @@
         os.makedirs(OUTPUT_DIR)
     written_files = []
 
-    gt_labels_4326_gdf = preapre_labels(SHPFILE, CATEGORY, supercategory=SUPERCATEGORY)
+    gt_labels_4326_gdf = prepare_labels(SHPFILE, CATEGORY, supercategory=SUPERCATEGORY)
 
     label_filepath = os.path.join(OUTPUT_DIR, 'labels.geojson')
     gt_labels_4326_gdf.to_file(label_filepath, driver='GeoJSON')

diff --git a/examples/swimming-pool-detection/GE/config_GE.yaml b/examples/swimming-pool-detection/GE/config_GE.yaml
@@ -45,7 +45,7 @@ train_model.py:
   detectron2_config_file: '../detectron2_config_GE.yaml' # path relative to the working_folder
   model_weights:
     model_zoo_checkpoint_url: "COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml"
-          
+
 make_detections.py:
   working_directory: output_GE
   log_subfolder: logs

diff --git a/examples/swimming-pool-detection/GE/detectron2_config_GE.yaml b/examples/swimming-pool-detection/GE/detectron2_config_GE.yaml
@@ -260,8 +260,6 @@ MODEL:
     LOSS_WEIGHT: 1.0
     NAME: SemSegFPNHead
     NORM: GN
-    NUM_CLASSES: 54
-  WEIGHTS: https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x/137260431/model_final_a54504.pkl
 OUTPUT_DIR: logs
 SEED: -1
 SOLVER:

diff --git a/examples/swimming-pool-detection/NE/config_NE.yaml b/examples/swimming-pool-detection/NE/config_NE.yaml
@@ -49,8 +49,8 @@ train_model.py:
       tst: COCO_tst.json
   detectron2_config_file: '../detectron2_config_NE.yaml' # path relative to the working_folder
   model_weights:
-    model_zoo_checkpoint_url: "COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml"
-
+    model_zoo_checkpoint_url: COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml
+        
 make_detections.py:
   working_directory: output_NE
   log_subfolder: logs

diff --git a/examples/swimming-pool-detection/NE/detectron2_config_NE.yaml b/examples/swimming-pool-detection/NE/detectron2_config_NE.yaml
@@ -261,12 +261,10 @@ MODEL:
     LOSS_WEIGHT: 1.0
     NAME: SemSegFPNHead
     NORM: GN
-    NUM_CLASSES: 54
-  WEIGHTS: https://dl.fbaipublicfiles.com/detectron2/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x/137260431/model_final_a54504.pkl
 OUTPUT_DIR: logs
 SEED: -1
 SOLVER:
-  BASE_LR: 0.005
+  BASE_LR: 0.00005
   BIAS_LR_FACTOR: 1.0
   CHECKPOINT_PERIOD: 1000
   CLIP_GRADIENTS:

diff --git a/helpers/detectron2.py b/helpers/detectron2.py
@@ -11,7 +11,8 @@
 
 from detectron2.engine.hooks import HookBase
 from detectron2.engine import DefaultTrainer
-from detectron2.data import build_detection_test_loader, DatasetMapper
+from detectron2.data import build_detection_test_loader, build_detection_train_loader, DatasetMapper
+from detectron2.data import transforms as T
 from detectron2.evaluation import COCOEvaluator
 from detectron2.utils import comm
 from detectron2.utils.logger import log_every_n_seconds
@@ -97,11 +98,8 @@ def build_evaluator(cls, cfg, dataset_name, output_folder=None):
 
     if output_folder is None:
         output_folder = os.path.join(cfg.OUTPUT_DIR, "inference")
-
-    os.makedirs("COCO_eval", exist_ok=True)
 
     return COCOEvaluator(dataset_name, None, False, output_folder)
-
 
   def build_hooks(self):
 
@@ -117,6 +115,20 @@ def build_hooks(self):
 
     return hooks
 
+
+class AugmentedCocoTrainer(CocoTrainer):
+
+  @classmethod
+  def build_train_loader(cls, cfg):
+      mapper = DatasetMapper(cfg, is_train=True, augmentations=[
+          T.RandomBrightness(0.5, 1.5),
+          T.RandomContrast(0.5, 1.5),
+          T.RandomSaturation(0.5, 1.5),
+          T.RandomLighting(0.5),
+          T.RandomFlip(),
+          T.ResizeShortestEdge(short_edge_length=[128, 1024], max_size=1333, sample_style='range')
+      ])
+      return build_detection_train_loader(cfg, mapper=mapper)
 
 
 # HELPER FUNCTIONS

diff --git a/helpers/functions_for_examples.py b/helpers/functions_for_examples.py
@@ -216,7 +216,7 @@ def get_tile_name(path, geom):
     return new_name
 
 
-def preapre_labels(labels_shp, category, supercategory):
+def prepare_labels(labels_shp, category, supercategory):
 
     ## Convert datasets shapefiles into geojson format
     logger.info('Convert labels shapefile into GeoJSON format (EPSG:4326)...')

diff --git a/helpers/misc.py b/helpers/misc.py
@@ -151,6 +151,16 @@ def clip_row(row, fact=fact):
 
 
 def format_logger(logger):
+    """
+    Configures the logger to format log messages with specific styles and colors based on their severity level.
+
+    Args:
+        logger (loguru.logger): The logger instance to be formatted.
+
+    Returns:
+        loguru.logger: The configured logger instance with custom formatting.
+
+    """
 
     logger.remove()
     logger.add(sys.stdout, format="{time:YYYY-MM-DD HH:mm:ss} - {level} - {message}",
@@ -166,6 +176,15 @@ def format_logger(logger):
 
 
 def find_category(df):
+    """
+    Ensures that the CATEGORY and SUPERCATEGORY columns are present in the input DataFrame.
+
+    Args:
+        df (pandas.DataFrame): DataFrame containing the GT labels.
+
+    Returns:
+        pandas.DataFrame: The input DataFrame with the CATEGORY and SUPERCATEGORY columns properly renamed.
+    """
 
     if 'category' in df.columns:
         df.rename(columns={'category': 'CATEGORY'}, inplace = True)

diff --git a/scripts/make_detections.py b/scripts/make_detections.py
@@ -14,7 +14,6 @@
 from tqdm import tqdm
 
 import geopandas as gpd
-import pandas as pd
 
 from detectron2.utils.logger import setup_logger
 setup_logger()