Datasets:

maysonma commited on
Commit
85a1d8a
1 Parent(s): 3bcc228

fix loader

Browse files
Files changed (1) hide show
  1. maplm_v2.py +38 -19
maplm_v2.py CHANGED
@@ -21,6 +21,7 @@ _CITATION = """\
21
  }
22
  """
23
 
 
24
  class MapLMBuilderConfig(datasets.BuilderConfig):
25
  """BuilderConfig for MapLM dataset."""
26
 
@@ -30,7 +31,6 @@ class MapLMBuilderConfig(datasets.BuilderConfig):
30
 
31
 
32
  class MapLMDataset(datasets.GeneratorBasedBuilder):
33
-
34
  BUILDER_CONFIG_CLASS = MapLMBuilderConfig
35
  BUILDER_CONFIGS = [
36
  MapLMBuilderConfig(
@@ -47,7 +47,7 @@ class MapLMDataset(datasets.GeneratorBasedBuilder):
47
  "images": datasets.Sequence(datasets.Value("string")),
48
  "question": datasets.Sequence(datasets.Value("string")),
49
  "options": datasets.Sequence(datasets.Sequence(datasets.Value("string"))),
50
- "answer": datasets.Sequence(datasets.Value("string")),
51
  "tag": datasets.Sequence(datasets.Value("string")),
52
  }
53
 
@@ -65,7 +65,7 @@ class MapLMDataset(datasets.GeneratorBasedBuilder):
65
  splits = []
66
  data_root = dl_manager.download("data/")
67
  for split in self.config.splits:
68
- annotation_file = os.path.join(data_root, split, f"{split}_v2.json")
69
  annotations = json.load(open(annotation_file))
70
  if split == "test":
71
  generator = datasets.SplitGenerator(
@@ -91,26 +91,45 @@ class MapLMDataset(datasets.GeneratorBasedBuilder):
91
  def _generate_examples(self, annotations):
92
  for i, anno_key in enumerate(annotations):
93
  data_item = {}
94
- data_item["frame_id"] = annotations[anno_key]["frame_id"]
95
- data_item["images"] = annotations[anno_key]["images"]
96
  data_item["question"] = []
97
  data_item["options"] = []
98
  data_item["answer"] = []
99
  data_item["tag"] = []
100
  for perception_key in annotations[anno_key]["QA"]["perception"]:
101
- data_item["question"].append(annotations[anno_key]["QA"]["perception"][perception_key]["question"])
102
- data_item["options"].append(annotations[anno_key]["QA"]["perception"][perception_key]["options"])
103
- data_item["answer"].append(annotations[anno_key]["QA"]["perception"][perception_key]["answer"])
104
- data_item["question_type"].append(annotations[anno_key]["QA"]["perception"][perception_key]["tag"])
105
-
106
- for behavior_key in annotations[anno_key]["QA"]["behavior"]:
107
- data_item["question"].append(annotations[anno_key]["QA"]["perception"][behavior_key]["question"])
108
- data_item["options"].append(annotations[anno_key]["QA"]["perception"][behavior_key]["options"])
109
- data_item["answer"].append(annotations[anno_key]["QA"]["perception"][behavior_key]["answer"])
110
- data_item["question_type"].append(annotations[anno_key]["QA"]["perception"][behavior_key]["tag"])
111
-
112
- yield i, data_item
113
-
 
 
 
 
 
 
114
 
 
 
 
 
 
 
 
 
 
 
 
 
 
115
 
116
-
 
21
  }
22
  """
23
 
24
+
25
  class MapLMBuilderConfig(datasets.BuilderConfig):
26
  """BuilderConfig for MapLM dataset."""
27
 
 
31
 
32
 
33
  class MapLMDataset(datasets.GeneratorBasedBuilder):
 
34
  BUILDER_CONFIG_CLASS = MapLMBuilderConfig
35
  BUILDER_CONFIGS = [
36
  MapLMBuilderConfig(
 
47
  "images": datasets.Sequence(datasets.Value("string")),
48
  "question": datasets.Sequence(datasets.Value("string")),
49
  "options": datasets.Sequence(datasets.Sequence(datasets.Value("string"))),
50
+ "answer": datasets.Sequence(datasets.Sequence(datasets.Value("string"))),
51
  "tag": datasets.Sequence(datasets.Value("string")),
52
  }
53
 
 
65
  splits = []
66
  data_root = dl_manager.download("data/")
67
  for split in self.config.splits:
68
+ annotation_file = os.path.join(data_root, f"{split}_v2.json")
69
  annotations = json.load(open(annotation_file))
70
  if split == "test":
71
  generator = datasets.SplitGenerator(
 
91
  def _generate_examples(self, annotations):
92
  for i, anno_key in enumerate(annotations):
93
  data_item = {}
94
+ data_item["frame_id"] = annotations[anno_key]["id"]
95
+ data_item["images"] = list(annotations[anno_key]["image_paths"].values())
96
  data_item["question"] = []
97
  data_item["options"] = []
98
  data_item["answer"] = []
99
  data_item["tag"] = []
100
  for perception_key in annotations[anno_key]["QA"]["perception"]:
101
+ data_item["question"].append(
102
+ annotations[anno_key]["QA"]["perception"][perception_key][
103
+ "question"
104
+ ]
105
+ )
106
+ data_item["options"].append(
107
+ annotations[anno_key]["QA"]["perception"][perception_key]["option"]
108
+ )
109
+ anno_answer = annotations[anno_key]["QA"]["perception"][perception_key][
110
+ "answer"
111
+ ]
112
+ if isinstance(anno_answer, list):
113
+ data_item["answer"].append(anno_answer)
114
+ else:
115
+ data_item["answer"].append([anno_answer])
116
+
117
+ data_item["tag"].append(
118
+ annotations[anno_key]["QA"]["perception"][perception_key]["tag"]
119
+ )
120
 
121
+ for behavior_key in annotations[anno_key]["QA"]["behavior"]:
122
+ data_item["question"].append(
123
+ annotations[anno_key]["QA"]["behavior"][behavior_key]["question"]
124
+ )
125
+ data_item["options"].append(
126
+ annotations[anno_key]["QA"]["behavior"][behavior_key]["option"]
127
+ )
128
+ data_item["answer"].append(
129
+ annotations[anno_key]["QA"]["behavior"][behavior_key]["answer"]
130
+ )
131
+ data_item["tag"].append(
132
+ annotations[anno_key]["QA"]["behavior"][behavior_key]["tag"]
133
+ )
134
 
135
+ yield i, data_item