diff --git a/megadetector/postprocessing/combine_batch_outputs.py b/megadetector/postprocessing/combine_batch_outputs.py
index 706eebc1..09f6acc0 100644
--- a/megadetector/postprocessing/combine_batch_outputs.py
+++ b/megadetector/postprocessing/combine_batch_outputs.py
@@ -27,6 +27,7 @@
 import argparse
 import sys
 import json
+
 from megadetector.utils import ct_utils
 
 
diff --git a/megadetector/postprocessing/generate_csv_report.py b/megadetector/postprocessing/generate_csv_report.py
index 3d8d915e..017a963d 100644
--- a/megadetector/postprocessing/generate_csv_report.py
+++ b/megadetector/postprocessing/generate_csv_report.py
@@ -335,7 +335,7 @@ def generate_csv_report(md_results_file,
 
                         # Classifications should always be sorted by confidence.  Not
                         # technically required, but always true in practice.
-                        assert is_list_sorted([c[1] for c in det['classifications']]), \
+                        assert is_list_sorted([c[1] for c in det['classifications']], reverse=True), \
                             'This script does not yet support unsorted classifications'
                         assert classification_category_id_to_name is not None, \
                             'If classifications are present, category mappings should be present'
diff --git a/megadetector/postprocessing/remap_detection_categories.py b/megadetector/postprocessing/remap_detection_categories.py
index 0b7f9a4b..d542aa84 100644
--- a/megadetector/postprocessing/remap_detection_categories.py
+++ b/megadetector/postprocessing/remap_detection_categories.py
@@ -98,29 +98,46 @@ def remap_detection_categories(input_file,
 
         # det = im['detections'][0]
         for det in im['detections']:
+
             input_category_id = det['category']
+
             if input_category_id not in input_category_id_to_output_category_id:
+
                 input_category_name = '[unknown]'
                 if input_category_id in input_categories:
                     input_category_name = input_categories[input_category_id]
+
                 s = 'Detection category {} ({}) not mapped'.format(
                     input_category_id,input_category_name)
+
                 if invalid_category_handling == 'error':
+
                     raise ValueError(s)
+
                 else:
+
                     assert invalid_category_handling == 'unknown'
                     print('Warning: {}'.format(s))
                     if 'unknown' in output_category_name_to_output_category_id:
                         output_category_id = output_category_name_to_output_category_id['unknown']
                     else:
-                        category_id_values = [int(x) for x in output_category_name_to_output_category_id.values()]
-                        unknown_category_id = max(category_id_values) + 1
+                        # Create a new "unknown" category for the unmapped category
+                        category_id_values = \
+                            [int(x) for x in output_category_name_to_output_category_id.values()]
+                        unknown_category_id = str(max(category_id_values) + 1)
                         output_category_name_to_output_category_id['unknown'] = unknown_category_id
+                        target_category_map[unknown_category_id] = 'unknown'
                         output_category_id = unknown_category_id
             else:
+
                 output_category_id = input_category_id_to_output_category_id[input_category_id]
+
+            # ...if this category ID has been mapped to the output dict
+
             det['category'] = output_category_id
 
+        # ...for each detection
+
     # ...for each image
 
     input_data['detection_categories'] = target_category_map