Add hdf5 support, regionprop options

scripts/cpn_inference.py

@@ -250,6 +250,8 @@ def main():
     parser.add_argument('--inputs_glob', default='*.*', type=str, help='Should `inputs` specify a directory, this'
                                                                        'can be used to filter for specific files '
                                                                        'in that directory.')
+    parser.add_argument('--inputs_method', default='imageio', help='Method used for loading non-hdf5 inputs.')
+    parser.add_argument('--inputs_dataset', default=None, help='Dataset name for hdf5 inputs.')
     parser.add_argument('-m', '--models', default='/models', help='Model directory or filename.')
     parser.add_argument('--devices', default='auto', type=str, help='Devices.')
     parser.add_argument('--accelerator', default='auto', type=str, help='Accelerator.')
@@ -279,6 +281,11 @@ def main():
                                                                    'Note: Intended for smaller images!')
     parser.add_argument('--truncated_images', action='store_true', help='Whether to support truncated images.')
     parser.add_argument('-p', '--properties', nargs='*', help='Region properties')
+    parser.add_argument('--spacing', default=1., type=float, help='The pixel spacing. Relevant for pixel-based '
+                                                                  'region properties.')
+    parser.add_argument('--separator', default='-', type=str,
+                        help='Separator string for region properties that are written to multiple columns. '
+                             'Default is "-" as in bbox-0, bbox-1, bbox-2, bbox-4.')
     args, unknown = parser.parse_known_args()
 
     if args.truncated_images:
@@ -320,9 +327,15 @@ def main():
     )
 
     for src in inputs:
-        dst = join(outputs, splitext(basename(src))[0] + '{ext}')
+        prefix, ext = splitext(basename(src))
+        dst = join(outputs, prefix + '{ext}')
         print(src, '-->', dst.format(ext='.*'), flush=True)
-        img = cd.load_image(src)
+        if ext in ('.h5', '.hdf5'):
+            assert args.inputs_dataset is not None, 'Please specify the dataset name for hdf5 inputs via --dataset <name>'
+            print('Read from h5:', args.inputs_dataset)
+            img = cd.from_h5(src, args.inputs_dataset)
+        else:
+            img = cd.load_image(src, method=args.inputs_method)
         y = cd.asnumpy(apply_model(
             img, models, trainer,
             crop_size=args.tile_size,
@@ -353,10 +366,12 @@ def main():
             if do_props:
                 if args.flat_labels:
                     assert flat_labels is not None
-                    cd.data.labels2property_table(flat_labels, props).to_csv(dst.format(ext='_flat.csv'))
+                    cd.data.labels2property_table(flat_labels, props, spacing=args.spacing,
+                                                  separator=args.separator).to_csv(dst.format(ext='_flat.csv'))
                 if args.labels or not args.flat_labels:
                     assert labels is not None
-                    cd.data.labels2property_table(labels, props).to_csv(dst.format(ext='.csv'))
+                    cd.data.labels2property_table(labels, props, spacing=args.spacing, separator=args.separator).to_csv(
+                        dst.format(ext='.csv'))
 
             if args.demo_figure:
                 from matplotlib import pyplot as plt