Datasets:
Anonymous commited on
Update nova.py
Browse files
nova.py
CHANGED
|
@@ -36,6 +36,9 @@ class NovaDataset(GeneratorBasedBuilder):
|
|
| 36 |
)
|
| 37 |
|
| 38 |
def _split_generators(self, dl_manager):
|
|
|
|
|
|
|
|
|
|
| 39 |
# Download CSVs
|
| 40 |
csvs = {
|
| 41 |
"captions": dl_manager.download("captions.csv"),
|
|
@@ -49,7 +52,7 @@ class NovaDataset(GeneratorBasedBuilder):
|
|
| 49 |
return [
|
| 50 |
SplitGenerator(
|
| 51 |
name=Split.TEST,
|
| 52 |
-
gen_kwargs={"
|
| 53 |
)
|
| 54 |
]
|
| 55 |
|
|
@@ -70,13 +73,14 @@ class NovaDataset(GeneratorBasedBuilder):
|
|
| 70 |
).to_dict()
|
| 71 |
meta = meta.set_index("case_id").to_dict("index")
|
| 72 |
|
|
|
|
| 73 |
|
| 74 |
for idx, row in captions.iterrows():
|
| 75 |
filename = row["filename"]
|
| 76 |
case_id = row["case_id"]
|
| 77 |
image_path = os.path.join(image_dir, filename)
|
| 78 |
record = {
|
| 79 |
-
"image":
|
| 80 |
"filename": filename,
|
| 81 |
"caption": row["caption"],
|
| 82 |
"case_id": case_id,
|
|
|
|
| 36 |
)
|
| 37 |
|
| 38 |
def _split_generators(self, dl_manager):
|
| 39 |
+
archive_path = dl_manager.download_and_extract("test/Images.zip")
|
| 40 |
+
image_files = list(dl_manager.iter_files(archive_path))
|
| 41 |
+
|
| 42 |
# Download CSVs
|
| 43 |
csvs = {
|
| 44 |
"captions": dl_manager.download("captions.csv"),
|
|
|
|
| 52 |
return [
|
| 53 |
SplitGenerator(
|
| 54 |
name=Split.TEST,
|
| 55 |
+
gen_kwargs={"image_files": image_files, "csvs": csvs},
|
| 56 |
)
|
| 57 |
]
|
| 58 |
|
|
|
|
| 73 |
).to_dict()
|
| 74 |
meta = meta.set_index("case_id").to_dict("index")
|
| 75 |
|
| 76 |
+
images_map = {os.path.basename(f): f for f in image_files}
|
| 77 |
|
| 78 |
for idx, row in captions.iterrows():
|
| 79 |
filename = row["filename"]
|
| 80 |
case_id = row["case_id"]
|
| 81 |
image_path = os.path.join(image_dir, filename)
|
| 82 |
record = {
|
| 83 |
+
"image": images_map.get(filename, None),
|
| 84 |
"filename": filename,
|
| 85 |
"caption": row["caption"],
|
| 86 |
"case_id": case_id,
|