Merge pull request #26 from swiss-territorial-data-lab/gs/improve_assessment

- Improve the assessment
- Make the validity check more versatile

Author: GwenaelleSa

helpers/metrics.py

@@ -54,10 +54,7 @@ def get_fractional_sets(dets_gdf, labels_gdf, iou_threshold=0.25):
     # IoU computation between labels and detections
     geom1 = candidates_tp_gdf['geometry'].to_numpy().tolist()
     geom2 = candidates_tp_gdf['label_geom'].to_numpy().tolist()
-    iou = []
-    for (i, ii) in zip(geom1, geom2):
-        iou.append(intersection_over_union(i, ii))
-    candidates_tp_gdf['IOU'] = iou
+    candidates_tp_gdf['IOU'] = [intersection_over_union(i, ii) for (i, ii) in zip(geom1, geom2)]
 
     # Filter detections based on IoU value
     best_matches_gdf = candidates_tp_gdf.groupby(['det_id'], group_keys=False).apply(lambda g:g[g.IOU==g.IOU.max()])
@@ -144,7 +141,7 @@ def get_metrics(tp_gdf, fp_gdf, fn_gdf, mismatch_gdf, id_classes=0):
             mismatched_fp_count = 0
             mismatched_fn_count = 0
         else:
-            mismatched_fp_count = len(tp_gdf[tp_gdf.det_class==id_cl])
+            mismatched_fp_count = len(mismatch_gdf[mismatch_gdf.det_class==id_cl])
             mismatched_fn_count = len(mismatch_gdf[mismatch_gdf.label_class==id_cl+1])
 
         fp_count = pure_fp_count + mismatched_fp_count

helpers/misc.py

@@ -55,9 +55,13 @@ def check_validity(poly_gdf, correct=False):
         if correct:
             print("Correction of the invalid geometries with the shapely function 'make_valid'...")
             invalid_poly = poly_gdf.loc[invalid_condition, 'geometry']
-            poly_gdf.loc[invalid_condition, 'geometry'] = [
-                make_valid(poly) for poly in invalid_poly
-                ]
+            try:
+                poly_gdf.loc[invalid_condition, 'geometry'] = [
+                    make_valid(poly) for poly in invalid_poly
+                    ]
+            except ValueError:
+                logger.info('Failed to fix geometries with "make_valid", try with a buffer of 0.')
+                poly_gdf.loc[invalid_condition, 'geometry'] = [poly.buffer(0) for poly in invalid_poly] 
         else:
             sys.exit(1)
 

scripts/assess_detections.py

@@ -392,6 +392,14 @@ def main(cfg_file_path):
         ].sort_values(by=['class', 'dataset']).to_csv(file_to_write, index=False)
         written_files.append(file_to_write)
 
+        tmp_df = metrics_by_cl_df[['dataset', 'TP_k', 'FP_k', 'FN_k']].groupby(by='dataset', as_index=False).sum()
+        tmp_df2 =  metrics_by_cl_df[['dataset', 'precision_k', 'recall_k']].groupby(by='dataset', as_index=False).mean()
+        global_metrics_df = tmp_df.merge(tmp_df2, on='dataset')
+
+        file_to_write = os.path.join(OUTPUT_DIR, 'global_metrics.csv')
+        global_metrics_df.to_csv(file_to_write, index=False)
+        written_files.append(file_to_write)
+
         # Save the confusion matrix
         na_value_category = tagged_dets_gdf.CATEGORY.isna()
         sorted_classes =  tagged_dets_gdf.loc[~na_value_category, 'CATEGORY'].sort_values().unique().tolist() + ['background']