Datasets:
Support streaming
#2
by
albertvillanova
HF staff
- opened
CC6204-Hackaton-Cub-Dataset.py
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
"""CC6204-Hackaton-Cub-Dataset: Multimodal"""
|
|
|
2 |
import os
|
3 |
import re
|
4 |
import datasets
|
@@ -102,14 +103,14 @@ class CubDataset(datasets.GeneratorBasedBuilder):
|
|
102 |
test_idx = []
|
103 |
|
104 |
# Download images
|
105 |
-
img_data_files = dl_manager.download_and_extract(_URLS["image_urls"])
|
106 |
-
text_data_files = dl_manager.download_and_extract(_URLS["text_urls"])
|
107 |
#logger.info(f"text_data_files: {text_data_files}")
|
108 |
#logger.info(f"text_data_files: {text_data_files[10]}")
|
109 |
|
110 |
|
111 |
-
img_path_files =
|
112 |
-
text_path_files =
|
113 |
|
114 |
for img, text in zip(img_path_files, text_path_files):
|
115 |
img_idx = _IMGNAME2ID[os.path.basename(img)]
|
|
|
1 |
"""CC6204-Hackaton-Cub-Dataset: Multimodal"""
|
2 |
+
import glob
|
3 |
import os
|
4 |
import re
|
5 |
import datasets
|
|
|
103 |
test_idx = []
|
104 |
|
105 |
# Download images
|
106 |
+
img_data_files = os.path.join(dl_manager.download_and_extract(_URLS["image_urls"]), "images") # skip _MACOSX dir
|
107 |
+
text_data_files = os.path.join(dl_manager.download_and_extract(_URLS["text_urls"]), "text") # skip _MACOSX dir
|
108 |
#logger.info(f"text_data_files: {text_data_files}")
|
109 |
#logger.info(f"text_data_files: {text_data_files[10]}")
|
110 |
|
111 |
|
112 |
+
img_path_files = sorted(glob.glob(os.path.join(img_data_files, "*", "*.jpg")))
|
113 |
+
text_path_files = sorted(glob.glob(os.path.join(text_data_files, "*", "*.txt")))
|
114 |
|
115 |
for img, text in zip(img_path_files, text_path_files):
|
116 |
img_idx = _IMGNAME2ID[os.path.basename(img)]
|