RoiSet facilitates object detection models

• created distinct generator class methods for RoiSet
• RoiSet.from_bounding_boxes can create RoiSet with just YXWH coordinates and no segmentation, e.g. as output from object detection models
• can also inspect and quantify overlap between and within RoiSet(s), at both bounding box and segmentation level
• can generate RoiSet from polygon segmentation e.g. in COCO segmentation models

model_server/base/roiset.py

@@ -310,7 +310,24 @@ class RoiSet(object):
         return RoiSet(acc_raw, df, params)
 
 
-    # TODO: add or overload for object detection case
+    # TODO: add a generator for the object detection case
+    @staticmethod
+    def from_bounding_boxes(
+        acc_raw: GenericImageDataAccessor,
+        yxhw_list: List,
+        params: RoiSetMetaParams = RoiSetMetaParams()
+    ):
+        df = pd.DataFrame([
+            {
+                'y0': yxhw[0],
+                'y1': yxhw[0] + yxhw[2],
+                'x0': yxhw[1],
+                'x1': yxhw[1] + yxhw[3],
+            } for yxhw in yxhw_list
+        ])
+        return RoiSet(acc_raw, df, params)
+
+
     @staticmethod
     def from_binary_mask(
             acc_raw: GenericImageDataAccessor,
@@ -705,7 +722,6 @@ class RoiSet(object):
             if k == 'annotated_zstacks':
                 record[k] = str(Path(k) / self.export_annotated_zstack(subdir, prefix=pr, **kp))
             if k == 'object_classes':
-                # for kc, acc in self.object_class_maps.items():
                 pr = 'classify_by_'
                 cnames = [c.split(pr)[1] for c in self._df.columns if c.startswith(pr)]
                 for n in cnames: