Add cats to hierarchy

bddl/data_generation/get_hierarchy_full.py

@@ -5,16 +5,30 @@
 problem that each synset can have multiple hypernyms
 '''
 import json
+import pandas as pd
 import pathlib
 from nltk.corpus import wordnet as wn
 
 HIERARCHY_OUTPUT_FN = pathlib.Path(__file__).parents[1] / "generated_data" / "output_hierarchy.json"
+CATEGORY_MAPPING_FN = pathlib.Path(__file__).parents[1] / "generated_data" / "category_mapping.csv"
+
+
+def add_igibson_objects(node, synset_to_cat):
+  '''
+  Go through the hierarchy and add the words associated with the synsets as attributes.
+  '''
+  # categories = 
+  if node["name"] in synset_to_cat:
+    node["igibson_categories"] = sorted(synset_to_cat[node["name"]])
+
+  if "children" in node:
+    for child_node in node["children"]:
+      add_igibson_objects(child_node, synset_to_cat)
 
 
 def add_path(path, hierarchy):
   if not path: 
     return 
-  # print("Path is nontrivial")
   synset = path[-1]
   if "children" not in hierarchy:
     hierarchy["children"] = []
@@ -56,6 +70,16 @@ def get_hierarchy(syn_prop_dict):
     for path in paths: 
       add_path(path[:-1], hierarchy)
 
+  synset_to_cat_raw = pd.read_csv(CATEGORY_MAPPING_FN)[["category", "synset"]].to_dict(orient="records")
+  synset_to_cat = {}
+  for rec in synset_to_cat_raw: 
+    syn, cat = rec["synset"], rec["category"]
+    if syn in synset_to_cat: 
+      synset_to_cat[syn].append(cat)
+    else:
+      synset_to_cat[syn] = [cat]
+  add_igibson_objects(hierarchy, synset_to_cat)
+
   with open(HIERARCHY_OUTPUT_FN, "w") as f:
     json.dump(hierarchy, f, indent=2)
   return hierarchy

bddl/data_generation/get_syn_prop_annots_canonical.py

@@ -83,7 +83,8 @@
     "nonSubstance": None,
     "sceneObject": None,
     "drapeable": None,
-    "physicalSubstance": None
+    "physicalSubstance": None,
+    "mixingTool": None,
 }
 
 ########### GET INPUT CANONICALS ###########

bddl/data_generation/process_prop_param_annots.py

@@ -10,8 +10,6 @@
 
 def add_cookable_params(propagated_canonical, props_to_syns, synset_nonexistent, param_but_no_prop, prop_but_no_param):
     # Cooking
-    print("COOKING")
-    print()
     cooking_params = pd.read_csv(PROP_PARAM_ANNOTS_DIR / "cooking.csv")
     cooking_params = cooking_params[cooking_params["require_cookable/overcookable_temperature"] == True][["synset", "value_cook(C)", "value_overcook(C)"]]
     for i, [synset, cook_temp, overcook_temp] in cooking_params.iterrows():
@@ -24,15 +22,12 @@ def add_cookable_params(propagated_canonical, props_to_syns, synset_nonexistent,
             prop_but_no_param["cookable"].append(synset)
             continue
         if synset not in propagated_canonical:
-            # print(f"{synset} not in syn-to-prop file.")
             synset_nonexistent["cookable"].append(synset)
             continue
         if "cookable" not in propagated_canonical[synset]:
-            # print(f"{synset} is not cookable but has cookable property param annotation.")
             param_but_no_prop["cookable"].append(synset)
             continue
         propagated_canonical[synset]["cookable"]["cook_temperature"] = float(cook_temp)
-        # syns_to_props[synset]["cookable"]["overcook_temperature"] = float(overcook_temp)
 
     for cookable_synset in props_to_syns["cookable"]:
         try: 
@@ -41,15 +36,11 @@ def add_cookable_params(propagated_canonical, props_to_syns, synset_nonexistent,
             continue
         if "cook_temperature" not in propagated_canonical[cookable_synset]["cookable"] and \
                         all(syn.name in propagated_canonical for syn in wn.synset(cookable_synset).hyponyms()):
-            # print(f"{cookable_synset} is annotated as cookable but does not have cookable property param annotation")
             prop_but_no_param["cookable"].append(cookable_synset)
 
 
 def add_coldsource_params(propagated_canonical, props_to_syns, synset_nonexistent, param_but_no_prop, prop_but_no_param):
     # coldSource
-    print(); print()
-    print("COLDSOURCE")
-    print()
     coldSource_params = pd.read_csv(PROP_PARAM_ANNOTS_DIR / "coldSource.csv")
     coldSource_params = coldSource_params[coldSource_params["require_coldsource_temperature"] == True][["synset", "value(C)"]]
     for i, [synset, temp] in coldSource_params.iterrows():
@@ -62,11 +53,9 @@ def add_coldsource_params(propagated_canonical, props_to_syns, synset_nonexisten
             prop_but_no_param["coldSource"].append(synset)
             continue
         if synset not in propagated_canonical:
-            # print(f"{synset} not in syn-to-prop file.")
             synset_nonexistent["coldSource"].append(synset)
             continue
         if "coldSource" not in propagated_canonical[synset]:
-            # print(f"{synset} is not coldSource but has coldSource property param annotation.")
             param_but_no_prop["coldSource"].append(synset)
             continue
         propagated_canonical[synset]["coldSource"]["temperature"] = float(temp)
@@ -79,36 +68,27 @@ def add_coldsource_params(propagated_canonical, props_to_syns, synset_nonexisten
             continue
         if "temperature" not in propagated_canonical[coldSource_synset]["coldSource"] and \
                         all(syn.name in propagated_canonical for syn in wn.synset(coldSource_synset).hyponyms()):
-            # print(f"{coldSource_synset} is annotated as coldSource_synset but does not have coldSource_synset property param annotation")
             prop_but_no_param["coldSource"].append(coldSource_synset)
 
 
 def add_heatsource_params(propagated_canonical, props_to_syns, synset_nonexistent, param_but_no_prop, prop_but_no_param):
     # heatSource
-    print(); print()
-    print("HEATSOURCE")
-    print()
     heatSource_params = pd.read_csv(PROP_PARAM_ANNOTS_DIR / "heatSource.csv")
 
     heatSource_params = heatSource_params[heatSource_params["require_heatsource_temperature"] == True][["synset", "value(C)"]]
-    # print(heatSource_params)
-    # import sys; sys.exit()
     for i, [synset, temp] in heatSource_params.iterrows():
         try:
             float(temp)
         except ValueError:
             prop_but_no_param["heatSource"].append(synset)
             continue
         if pd.isna(float(temp)):
-            print(synset)
             prop_but_no_param["heatSource"].append(synset)
             continue
         if synset not in propagated_canonical:
-            # print(f"{synset} not in syn-to-prop file.")
             synset_nonexistent["heatSource"].append(synset)
             continue
         if "heatSource" not in propagated_canonical[synset]:
-            # print(f"{synset} is not heatSource but has heatSource property param annotation.")
             param_but_no_prop["heatSource"].append(synset)
             continue
         propagated_canonical[synset]["heatSource"]["temperature"] = float(temp)
@@ -121,7 +101,6 @@ def add_heatsource_params(propagated_canonical, props_to_syns, synset_nonexisten
             continue
         if "temperature" not in propagated_canonical[heatSource_synset]["heatSource"] and \
                         all(syn.name in propagated_canonical for syn in wn.synset(heatSource_synset).hyponyms()):        
-            # print(f"{heatSource_synset} is annotated as heatSource but does not have heatSource property param annotation")
             prop_but_no_param["heatSource"].append(heatSource_synset)
 
 # TODO check if that everything that SHOULD have a certain parameter annotation DOES have it 

bddl/data_generation/prop_config.py

@@ -33,5 +33,6 @@
     "particleSource",
     "needsOrientation",
     "waterCook",
+    "mixingTool",
     "sceneObject"
 ]