Learning/ObjectBased.py

@@ -93,7 +93,7 @@ class ObjectBasedClassifier:
         assert('folds' in self.training_base.keys())
         models = []
         results = []
-        yt_yp = []
+        truelabs = np.array([])
         for tr_i, ts_i in tqdm(self.training_base['folds'], desc='Training'):
             models.append(RandomForestClassifier(n_estimators=n_estimators))
             models[-1].fit(self.training_base['X'][tr_i], self.training_base[class_field][tr_i])
@@ -102,6 +102,7 @@ class ObjectBasedClassifier:
             c = np.delete(c, np.isin(l, self.training_base['dummy_ids']))
             l = np.delete(l, np.isin(l, self.training_base['dummy_ids']))
             y_true, y_pred = self.obia_base.true_pred_bypixel(l, c, class_field)
+            truelabs = np.unique(np.concatenate((truelabs,y_true,y_pred)))
             results.append(
                 {
                     'conf_matrix': confusion_matrix(y_true, y_pred, labels=np.unique(self.training_base[class_field])),
@@ -127,7 +128,8 @@ class ObjectBasedClassifier:
             'f1_mean': np.mean([x['p_r_f1'][2] for x in results], axis=0),
             'f1_std': np.std([x['p_r_f1'][2] for x in results], axis=0),
             'importance_mean': {k:v for k, v in zip(self.obia_base.get_vars(), np.mean(all_imp, axis=0))},
-            'importance_std': {k:v for k, v in zip(self.obia_base.get_vars(), np.std(all_imp, axis=0))}
+            'importance_std': {k:v for k, v in zip(self.obia_base.get_vars(), np.std(all_imp, axis=0))},
+            'actual_labels': list(truelabs)
         }
         return models, summary, results
 

OBIA/OBIABase.py

@@ -92,9 +92,12 @@ class OBIABase:
         ref_ol.Execute()
         self.ref_obj_layer_pipe = [in_seg, ras_id, ref_ol]
 
+        self.ref_db = None
+        '''
         self.ref_db = pd.DataFrame(data=[],
                                    columns=['area', 'orig_label', 'polygon_id'] + class_field,
                                    index=[])
+        '''
         r = otb.itkRegion()
         for tn, t in tqdm(self.tiles.items(), desc='Init. Ref. DB', total=len(self.tiles)):
             r['index'][0], r['index'][1] = t[0], t[1]
@@ -104,14 +107,19 @@ class OBIABase:
             tile_ref_ol = ref_ol.GetImageAsNumpyArray('out').astype(np.int32)
             tile_int_img = intensity_img.GetVectorImageAsNumpyArray('out').astype(int)
             rp = regionprops(tile_ref_ol, intensity_image=tile_int_img)
-            self.ref_db = pd.concat([
-                self.ref_db,
-                pd.DataFrame(
-                    data=[np.insert(o.intensity_min, 0, o.area) for o in rp if self.obj_to_tile[o.label] == tn],
-                    columns=self.ref_db.columns,
-                    index=[o.label for o in rp if self.obj_to_tile[o.label] == tn]
-                )]
-            )
+            if self.ref_db is None:
+                self.ref_db = pd.DataFrame(data=[np.insert(o.intensity_min, 0, o.area) for o in rp if self.obj_to_tile[o.label] == tn],
+                                           columns=['area', 'orig_label', 'polygon_id'] + class_field,
+                                           index=[o.label for o in rp if self.obj_to_tile[o.label] == tn])
+            else:
+                self.ref_db = pd.concat([
+                    self.ref_db,
+                    pd.DataFrame(
+                        data=[np.insert(o.intensity_min, 0, o.area) for o in rp if self.obj_to_tile[o.label] == tn],
+                        columns=self.ref_db.columns,
+                        index=[o.label for o in rp if self.obj_to_tile[o.label] == tn]
+                    )]
+                )
 
         return
 

Postprocessing/MapFormatting.py

 import os
 
-def parse_colormap_file(fn):
+def parse_colormap_file(fn, filter=None):
     labels = []
     colors = []
     class_names = []
@@ -8,6 +8,8 @@ def parse_colormap_file(fn):
     with open(fn, 'r') as f:
         for l in f.read().splitlines():
             sl = l.split(' ')
+            if filter is not None and int(sl[0]) not in filter:
+                continue
             labels.append(int(sl[0]))
             colors.append((int(sl[1]),int(sl[2]),int(sl[3]),int(sl[4])))
             class_names.append(' '.join(sl[5:]))

Postprocessing/Report.py

@@ -30,12 +30,12 @@ def filter_and_order_importance(summary, importance_perc, max_num_var=35):
 
 def generate_report_figures(map, palette_fn, results, summary, out_dir, map_name=None,
                             importance_perc=0.75, max_variables=35):
-    labels, class_names, colors = parse_colormap_file(palette_fn)
+    labels, class_names, colors = parse_colormap_file(palette_fn, filter=summary['actual_labels'])
     colors_norm = [(x[0]/255,x[1]/255,x[2]/255,x[3]/255) for x in colors]
     with plt.ioff():
-        #font = {'weight': 'normal',
-        #        'size': 8}
-        #plt.rc('font', **font)
+        font = {'weight': 'normal',
+                'size': 8}
+        plt.rc('font', **font)
         if not os.path.exists(out_dir):
             os.makedirs(out_dir)
         if not isinstance(results, list):
@@ -73,7 +73,7 @@ def generate_report_figures(map, palette_fn, results, summary, out_dir, map_name
         plt.tight_layout()
         plt.savefig(of['summary'], dpi=300)
 
-        imp_m, imp_s, imp_n = filter_and_order_importance(summary, importance_perc)
+        imp_m, imp_s, imp_n = filter_and_order_importance(summary, importance_perc, max_num_var=max_variables)
 
         fig, ax = plt.subplots()
         ax.barh(range(len(imp_n)), imp_m, xerr=imp_s, align='center')
@@ -194,7 +194,7 @@ def generate_pdf(of, out_pdf, name='output'):
 def generate_text_report(results, summary, palette_fn, output_fn, name='output'):
     if os.path.dirname(output_fn) != '':
         os.makedirs(os.path.dirname(output_fn), exist_ok=True)
-    labels, class_names, _ = parse_colormap_file(palette_fn)
+    labels, class_names, _ = parse_colormap_file(palette_fn, filter=summary['actual_labels'])
     lines = ['MORINGA Final Report for chain {}, {}'.format(name, datetime.now().strftime('%Y-%m-%d %Hh%M')), '']
     
     table_lines = []

conda_recipe.txt

+conda create --name moringa-env -y python=3.10.13
+conda activate moringa-env
+conda install -y -c conda-forge mamba
+mamba install -y -c conda-forge gdal=3.4.2 pyproj rasterio geopandas shapely scikit-learn scikit-image
+pip install psutil planetary_computer fpdf2 planet eodag tabulate
+