Multiresolution flag (#215)

* Added target measures to custom tutorial

* Fix nbval issues

* Added multiresolution flag

btk/draw_blends.py

@@ -172,6 +172,7 @@ def __init__(
             self.surveys = surveys
         else:
             raise TypeError("surveys must be a Survey object or a list of Survey objects.")
+        self.is_multiresolution = len(self.surveys) > 1
 
         self.stamp_size = stamp_size
         self.add_noise = add_noise
@@ -268,7 +269,7 @@ def __next__(self):
                         format="ascii",
                         overwrite=True,
                     )
-        if len(self.surveys) > 1:
+        if self.is_multiresolution:
             output = {
                 "blend_images": blend_images,
                 "isolated_images": isolated_images,

btk/measure.py

@@ -89,7 +89,9 @@ def add_pixel_columns(catalog, wcs):
     return catalog_t
 
 
-def basic_measure(batch, idx, channels_last=False, surveys=None, **kwargs):
+def basic_measure(
+    batch, idx, channels_last=False, surveys=None, is_multiresolution=False, **kwargs
+):
     """Return centers detected with skimage.feature.peak_local_max.
 
     NOTE: If this function is used with the multiresolution feature,
@@ -106,7 +108,7 @@ def basic_measure(batch, idx, channels_last=False, surveys=None, **kwargs):
     channel_indx = 0 if not channels_last else -1
 
     # multiresolution
-    if isinstance(batch["blend_images"], dict):
+    if is_multiresolution:
         if surveys is None:
             raise ValueError("surveys are required in order to use the MR feature.")
         surveys = kwargs.get("surveys", None)
@@ -129,7 +131,15 @@ def basic_measure(batch, idx, channels_last=False, surveys=None, **kwargs):
     return {"catalog": catalog}
 
 
-def sep_measure(batch, idx, channels_last=False, surveys=None, sigma_noise=1.5, **kwargs):
+def sep_measure(
+    batch,
+    idx,
+    channels_last=False,
+    surveys=None,
+    sigma_noise=1.5,
+    is_multiresolution=False,
+    **kwargs,
+):
     """Return detection, segmentation and deblending information with SEP.
 
     NOTE: If this function is used with the multiresolution feature,
@@ -148,7 +158,7 @@ def sep_measure(batch, idx, channels_last=False, surveys=None, sigma_noise=1.5,
     channel_indx = 0 if not channels_last else -1
 
     # multiresolution
-    if isinstance(batch["blend_images"], dict):
+    if is_multiresolution:
         if surveys is None:
             raise ValueError("surveys are required in order to use the MR feature.")
         survey_name = surveys[0].name
@@ -184,7 +194,7 @@ def sep_measure(batch, idx, channels_last=False, surveys=None, sigma_noise=1.5,
     t["ra"], t["dec"] = wcs.pixel_to_world_values(catalog["x"], catalog["y"])
 
     # If multiresolution, return only the catalog
-    if isinstance(batch["blend_images"], dict):
+    if is_multiresolution:
         return {"catalog": t}
     else:
         return {
@@ -241,6 +251,7 @@ def __init__(
         self.batch_size = self.draw_blend_generator.batch_size
         self.channels_last = self.draw_blend_generator.channels_last
         self.surveys = self.draw_blend_generator.surveys
+        self.is_multiresolution = self.draw_blend_generator.is_multiresolution
         self.verbose = verbose
         self.save_path = save_path
 
@@ -249,6 +260,7 @@ def __init__(
         for m in self.measure_kwargs:
             m["channels_last"] = self.channels_last
             m["surveys"] = self.surveys
+            m["is_multiresolution"] = self.is_multiresolution
 
     def __iter__(self):
         """Return iterator which is the object itself."""
@@ -356,7 +368,7 @@ def __next__(self):
                     )
                 # If multiresolution, we reverse the order between the survey name and
                 # the index of the blend
-                if isinstance(blend_output["blend_list"], dict):
+                if self.is_multiresolution:
                     survey_keys = list(blend_output["blend_list"].keys())
                     # We duplicate the catalog for each survey to get the pixel coordinates
                     catalogs_temp = {}

btk/metrics.py

@@ -724,6 +724,7 @@ def __init__(
         self.f_distance = f_distance
         self.distance_threshold_match = distance_threshold_match
         self.verbose = verbose
+        self.is_multiresolution = self.measure_generator.is_multiresolution
 
     def __next__(self):
         """Returns metric results calculated on one batch."""
@@ -732,7 +733,7 @@ def __next__(self):
 
         metrics_results = {}
         for meas_func in measure_results["catalog"].keys():
-            if isinstance(blend_results["isolated_images"], dict):
+            if self.is_multiresolution:
                 metrics_results_f = {}
                 for i, surv in enumerate(blend_results["isolated_images"].keys()):
                     additional_params = {

notebooks/scarlet-measure.ipynb

@@ -88,30 +88,28 @@
    },
    "outputs": [],
    "source": [
-    "def scarlet_measure(batch,idx,channels_last=False,**kwargs):\n",
+    "def scarlet_measure(batch,idx,channels_last=False, is_multiresolution=False,**kwargs):\n",
     "    \"\"\"Measure function for SCARLET\n",
     "    \"\"\"\n",
     "    sigma_noise = kwargs.get(\"sigma_noise\", 1.5)    \n",
     "    surveys = kwargs.get(\"surveys\", None)\n",
     "    \n",
     "    #Fist we carry out the detection, using SExtractor (sep being the python implementation)\n",
     "    # We need to differentiate between the multiresolution and the regular case\n",
-    "    if isinstance(batch[\"blend_images\"], dict):\n",
+    "    if is_multiresolution:\n",
     "        survey_name = surveys[0].name\n",
     "        image = batch[\"blend_images\"][survey_name][idx]\n",
     "        # Put the image in the channels first format if not already the case\n",
     "        image = np.moveaxis(image,-1,0) if channels_last else image\n",
     "        coadd = np.mean(image, axis=0)\n",
     "        wcs_ref = batch[\"wcs\"][survey_name]\n",
     "        psf = np.array([p.drawImage(galsim.Image(image.shape[1],image.shape[2]),scale=surveys[0].pixel_scale).array for p in batch[\"psf\"][survey_name]])\n",
-    "        multiresolution = True\n",
     "    else:\n",
     "        image = batch[\"blend_images\"][idx]\n",
     "        image = np.moveaxis(image,-1,0) if channels_last else image\n",
     "        coadd = np.mean(image, axis=0)\n",
     "        psf = np.array([p.drawImage(galsim.Image(image.shape[1],image.shape[2]),scale=surveys[0].pixel_scale).array for p in batch[\"psf\"]])\n",
     "        wcs_ref = batch[\"wcs\"]\n",
-    "        multiresolution = False\n",
     "    stamp_size = coadd.shape[0]\n",
     "    \n",
     "    bkg = sep.Background(coadd)\n",
@@ -123,7 +121,7 @@
     "    if len(catalog) == 0:\n",
     "        t = astropy.table.Table()\n",
     "        t[\"ra\"], t[\"dec\"] = wcs_ref.pixel_to_world_values(catalog[\"x\"], catalog[\"y\"])\n",
-    "        if multiresolution:\n",
+    "        if is_multiresolution:\n",
     "            return {\"catalog\":t,\"segmentation\":None,\"deblended_images\":{s.name: np.array([np.zeros((len(s.filters),batch[\"blend_images\"][s.name][idx].shape[1],\n",
     "                                                batch[\"blend_images\"][s.name][idx].shape[1]))]) for s in surveys}}\n",
     "        else:\n",
@@ -200,7 +198,7 @@
     "    except AssertionError: #If the fitting fails\n",
     "        t = astropy.table.Table()\n",
     "        t[\"ra\"], t[\"dec\"] = wcs_ref.pixel_to_world_values(catalog[\"x\"], catalog[\"y\"])\n",
-    "        if multiresolution:\n",
+    "        if is_multiresolution:\n",
     "            deblended_images={s.name: np.array([np.zeros((len(s.filters),batch[\"blend_images\"][s.name][idx].shape[1],\n",
     "                                              batch[\"blend_images\"][s.name][idx].shape[1])) for c in catalog]) for s in surveys}\n",
     "        else:\n",