diff --git a/fixtracks/utils/trackingdata.py b/fixtracks/utils/trackingdata.py
index 4dcc3f0..213b520 100644
--- a/fixtracks/utils/trackingdata.py
+++ b/fixtracks/utils/trackingdata.py
@@ -14,8 +14,8 @@ class TrackingData(QObject):
         if "userlabeled" not in self._data.keys():
             self._data["userlabeled"] = np.zeros_like(self["frame"], dtype=bool)
         self._columns = [k for k in self._data.keys()]
-        self._indices = self["index"]
-        self._selection = np.asarray([])
+        self._indices = np.arange(len(self["index"]), dtype=int)
+        self._selection_indices = np.asarray([])
         self._selected_ids = None
 
     @property
@@ -43,8 +43,8 @@ class TrackingData(QObject):
         ids = np.sort(ids)
         indexes = np.ones_like(ids, dtype=int) * -1
         j = 0
-        for idx, i in enumerate(self._indices):
-            if i == ids[j]:
+        for idx in self._indices:
+            if self["index"][idx] == ids[j]:
                 indexes[j] = idx
                 j += 1
             if j == len(indexes):
@@ -54,19 +54,23 @@ class TrackingData(QObject):
 
     @property
     def selectionIndices(self):
-        return self._selection
+        return self._selection_indices
+
+    @property
+    def selectionIDs(self):
+        return self._selected_ids
 
     def setSelectionRange(self, col, start, stop):
         logging.info("Trackingdata: set selection range based on column %s to %.2f - %.2f", col, start, stop)
         col_indices = np.where((self[col] >= start) & (self[col] < stop))[0]
-        self._selection = self._indices[col_indices]
+        self._selection_indices = self._indices[col_indices]
         if len(col_indices) < 1:
             logging.warning("TrackingData: Selection range is empty!")
 
     def selectedData(self, col:str):
         if col not in self.columns:
             logging.error("TrackingData:selectedData: Invalid column name! %s", col)
-        return self[col][self._selection]
+        return self[col][self._selection_indices]
 
     def setSelection(self, ids):
         """
@@ -78,8 +82,9 @@ class TrackingData(QObject):
             An array-like object containing the IDs to be set as user selections. 
         """
         logging.debug("TrackingData.setSelection: %i number of ids", len(ids))
-        self._selection = self._find(ids)
+        self._selection_indices = self._find(ids)
         self._selected_ids = ids
+        print(self._selected_ids, self._selection_indices)
 
     def setTrack(self, track_id:int, setUserLabeled:bool=True)-> None:
         """Assign a new track_id to the user-selected detections
@@ -92,9 +97,12 @@ class TrackingData(QObject):
             Should the "userlabeled" state of the detections be set to True? Otherwise they will be left untouched.
         """
         logging.info("TrackingData: set track id %i for selection, set user-labeled status %s", track_id, str(setUserLabeled))
-        self["track"][self._selection] = track_id
+        print(self._selected_ids, self._selection_indices)
+        print("before: ", self["track"][self._selection_indices], self["frame"][self._selection_indices])
+        self["track"][self._selection_indices] = track_id
         if setUserLabeled:
             self.setUserLabeledStatus(True, True)
+        print("after: ", self["track"][self._selection_indices], self["frame"][self._selection_indices])
 
     def setUserLabeledStatus(self, new_status: bool, selection=True):
         """Sets the status of the "userlabeled" column to a given value (True|False). This can done for ALL data in one go, or only for the UserSelection.
@@ -109,7 +117,7 @@ class TrackingData(QObject):
         logging.debug("TrackingData: (Re-)setting assignment status of %s to %s", 
                       "user selected data" if selection else " ALL", str(new_status))
         if selection:
-            self["userlabeled"][self._selection] = new_status
+            self["userlabeled"][self._selection_indices] = new_status
         else:
             self["userlabeled"][:] = new_status
 
@@ -123,12 +131,14 @@ class TrackingData(QObject):
 
     def deleteDetections(self, ids=None):
         if ids is not None:
+            logging.debug("TrackingData.deleteDetections of %i detections", len(ids))
             del_indices = self._find(ids)
         else:
-            del_indices = self._indices
+            logging.debug("TrackingData.deleteDetections of all selected detections (%i)", len(self._selected_ids))
+            del_indices = self._selected_ids
         for c in self._columns:
             self._data[c] = np.delete(self._data[c], del_indices, axis=0)
-        self._indices = self["index"]
+        self._indices = self._indices[:-len(del_indices)]
         self._selected_ids = np.setdiff1d(self._selected_ids, del_indices)
 
     def assignTracks(self, tracks:np.ndarray):
@@ -171,10 +181,10 @@ class TrackingData(QObject):
                         and M is number of keypoints
         """
         if selection:
-            if len(self._selection) < 1:
+            if len(self._selection_indices) < 1:
                 logging.info("TrackingData.coordinates returns empty array, not detections in range!")
                 return np.ndarray([])
-            return np.stack(self["keypoints"][self._selection]).astype(np.float32)
+            return np.stack(self["keypoints"][self._selection_indices]).astype(np.float32)
         return np.stack(self["keypoints"]).astype(np.float32)
 
     def keypointScores(self, selection=False):
@@ -188,10 +198,10 @@ class TrackingData(QObject):
             with N the number of detections and M the number of keypoints.
         """
         if selection:
-            if len(self._selection) < 1:
+            if len(self._selection_indices) < 1:
                 logging.info("TrackingData.scores returns empty array, not detections in range!")
                 return None
-            return np.stack(self["keypoint_score"][self._selection]).astype(np.float32)
+            return np.stack(self["keypoint_score"][self._selection_indices]).astype(np.float32)
         return np.stack(self["keypoint_score"]).astype(np.float32)
 
     def centerOfGravity(self, selection=False, threshold=0.8, nodes=[0,1,2]):