integrate-new-concepts (#3)

- new concepts columns (8a41a50b2140bd507cb9e698114b86dabf407c5b)


Co-authored-by: Yoann Poupart <Xmaster6y@users.noreply.huggingface.co>

src/constants.py

@@ -5,7 +5,7 @@ import os
 import pathlib
 
 
-DATASET_NAME = "Xmaster6y/fruit-vegetable-concepts"
+DATASET_NAME = "mulsi/fruit-vegetable-concepts"
 CONCEPTS = [
     # Shapes
     "sphere",

src/global_variables.py

@@ -78,7 +78,8 @@ def update_votes(
         all_votes[s_id] = {}
     all_votes[s_id][username] = {c: c in voted_concepts for c in CONCEPTS}
     new_concepts = compute_concepts(all_votes[s_id])
-    all_metadata[current_split][idx]["concepts"] = new_concepts
+    for concept, concept_value in new_concepts.items():
+        all_metadata[current_split][idx][concept] = concept_value
     
 def compute_concepts(votes):
     vote_sum = {c: 0 for c in CONCEPTS}
@@ -122,7 +123,7 @@ def save_current_work(
             for row in reader:
                 s_id = row["id"]
                 if s_id in all_votes:
-                    row["concepts"] = compute_concepts(all_votes[s_id])
+                    row.update(compute_concepts(all_votes[s_id]))
                 new_metadata[split].append(row)
         with jsonlines.open(f"{ASSETS_FOLDER}/{DATASET_NAME}/data/{split}/metadata.jsonl", mode='w') as writer:
             writer.write_all(new_metadata[split])

src/label_interface.py

@@ -12,7 +12,7 @@ from src.constants import CONCEPTS, ASSETS_FOLDER, DATASET_NAME
 
 
 def filter_sample(sample, concepts, username, sample_type):
-    has_concepts = all([sample["concepts"].get(c, False) for c in concepts])
+    has_concepts = all([sample[c] for c in concepts])
     if not has_concepts:
         return False
     if "votes" in sample and username in sample["votes"]:
@@ -53,14 +53,14 @@ def get_next_image(
         except KeyError:
             voted_concepts = []
             unseen_concepts = []
-        tie_concepts = [c for c in sample["concepts"] if sample["concepts"][c] is None]
+        tie_concepts = [c for c in CONCEPTS if sample[c] is None]
 
         return (
             image_path, 
             voted_concepts, 
             f"{split}:{sample_idx}", 
             sample["class"], 
-            sample["concepts"],
+            {c: sample[c] for c in CONCEPTS},
             unseen_concepts,
             tie_concepts,
             filtered_indices, 

src/sample_interface.py

@@ -37,14 +37,14 @@ def get_image(
     except KeyError:
         voted_concepts = []
         unseen_concepts = []
-    tie_concepts = [c for c in sample["concepts"] if sample["concepts"][c] is None]
+    tie_concepts = [c for c in CONCEPTS if sample[c] is None]
 
     return (
         image_path, 
         voted_concepts, 
         f"{split}:{sample_idx}", 
         sample["class"], 
-        sample["concepts"],
+        {c: sample[c] for c in CONCEPTS},
         str(sample_idx),
         unseen_concepts,
         tie_concepts,