ydshieh commited on
Commit ·
8ffa189
1
Parent(s): 9628bd8
Add dataset usage example script
Browse files- dataset_example.py +26 -0
dataset_example.py
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from datasets import load_dataset
|
| 2 |
+
from PIL import Image
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
dataset_name = "image_caption_dataset.py"
|
| 6 |
+
dataset_config_name = "coco_2017"
|
| 7 |
+
cache_dir = None
|
| 8 |
+
keep_in_memory = False
|
| 9 |
+
data_dir = "./"
|
| 10 |
+
|
| 11 |
+
dataset = load_dataset(
|
| 12 |
+
dataset_name, dataset_config_name, cache_dir=cache_dir, keep_in_memory=keep_in_memory, data_dir=data_dir
|
| 13 |
+
)
|
| 14 |
+
|
| 15 |
+
for example in dataset["train"]:
|
| 16 |
+
print(example)
|
| 17 |
+
# with Image.open(example['image_file']) as image:
|
| 18 |
+
# image.show()
|
| 19 |
+
break
|
| 20 |
+
|
| 21 |
+
for _idx, example in enumerate(dataset["validation"]):
|
| 22 |
+
if _idx >= 5:
|
| 23 |
+
break
|
| 24 |
+
print(example)
|
| 25 |
+
with Image.open(example['image_file']) as image:
|
| 26 |
+
image.show()
|