1) RRandomCrop Runtime Error Fixed (when only one bounding box exists in the cropped); 2) mAP Print Error Fixed (when #class of model > # class of data); 3) confusion matrix color-theme error fixed!

mmrotate/core/evaluation/eval_map.py

@@ -276,12 +276,20 @@ def print_map_summary(mean_ap,
     if scale_ranges is not None:
         assert len(scale_ranges) == num_scales
 
-    num_classes = len(results)
+    # num_classes = len(results)
+    # recalls = np.zeros((num_scales, num_classes), dtype=np.float32)
+
+    ####################################################################################################################
+    num_classes = len(dataset) if dataset else len(results)
+    num_results_classes = len(results)
+    min_classes = min(num_classes, num_results_classes)
+    recalls = np.zeros((num_scales, min_classes), dtype=np.float32)
+    ####################################################################################################################
 
-    recalls = np.zeros((num_scales, num_classes), dtype=np.float32)
     aps = np.zeros((num_scales, num_classes), dtype=np.float32)
     num_gts = np.zeros((num_scales, num_classes), dtype=int)
-    for i, cls_result in enumerate(results):
+    # for i, cls_result in enumerate(results):
+    for i, cls_result in enumerate(results[:min_classes]):
         if cls_result['recall'].size > 0:
             recalls[:, i] = np.array(cls_result['recall'], ndmin=2)[:, -1]
         aps[:, i] = cls_result['ap']
@@ -290,7 +298,8 @@ def print_map_summary(mean_ap,
     if dataset is None:
         label_names = [str(i) for i in range(num_classes)]
     else:
-        label_names = dataset
+        # label_names = dataset
+        label_names = dataset[:min_classes]
 
     if not isinstance(mean_ap, list):
         mean_ap = [mean_ap]
@@ -302,11 +311,13 @@ def print_map_summary(mean_ap,
         table_data = [header]
         for j in range(num_classes):
             row_data = [
-                label_names[j], num_gts[i, j], results[j]['num_dets'],
+                # label_names[j], num_gts[i, j], results[j]['num_dets'],
+                label_names[j], num_gts[i, j], results[j].get('num_dets', 0),
                 f'{recalls[i, j]:.3f}', f'{aps[i, j]:.3f}'
             ]
             table_data.append(row_data)
         table_data.append(['mAP', '', '', '', f'{mean_ap[i]:.3f}'])
         table = AsciiTable(table_data)
         table.inner_footing_row_border = True
         print_log('\n' + table.table, logger=logger)
+

mmrotate/datasets/pipelines/transforms.py

@@ -352,7 +352,7 @@ def _crop_data(self, results, crop_size, allow_negative_crop):
             img = img[crop_y1:crop_y2, crop_x1:crop_x2, ...]
             img_shape = img.shape
             results[key] = img
-        results['img_shape'] = img_shape
+        results['img_shape'] = img_shape[:2]
 
         height, width, _ = img_shape
 
@@ -375,6 +375,8 @@ def _crop_data(self, results, crop_size, allow_negative_crop):
             if (key == 'gt_bboxes' and not valid_inds.any()
                     and not allow_negative_crop):
                 return None
+
+            valid_inds = np.atleast_1d(valid_inds)
             results[key] = bboxes[valid_inds, :]
             # label fields. e.g. gt_labels and gt_labels_ignore
             label_key = self.bbox2label.get(key)

tools/analysis_tools/confusion_matrix.py

@@ -226,6 +226,8 @@ def plot_confusion_matrix(confusion_matrix,
     if save_dir is not None:
         plt.savefig(
             os.path.join(save_dir, 'confusion_matrix.png'), format='png')
+        plt.savefig(
+            os.path.join(save_dir, 'confusion_matrix.svg'), format='svg')
     if show:
         plt.show()
 
@@ -261,8 +263,10 @@ def main():
         confusion_matrix,
         dataset.CLASSES + ('background', ),
         save_dir=args.save_dir,
-        show=args.show)
+        show=args.show,
+        color_theme=args.color_theme)
 
 
 if __name__ == '__main__':
     main()
+