File size: 3,511 Bytes
808187d
 
 
 
 
 
 
eadf52e
 
808187d
f7f90b0
62e9177
808187d
51e6200
e774f45
 
808187d
3371f7f
808187d
 
 
 
 
 
e774f45
 
808187d
 
 
d11e825
1ba6734
 
808187d
5c728e6
 
d11e825
808187d
62e9177
808187d
 
 
d11e825
 
e774f45
d11e825
 
15fe087
d11e825
 
 
9d9dbc4
d11e825
e774f45
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
353c340
e774f45
 
 
 
 
 
 
 
 
 
 
90ab190
 
 
 
e774f45
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
"""CC6204-Hackaton-Cub-Dataset: Multimodal"""
import os
import re
import datasets

import pandas as pd

from requests import get

logger = datasets.logging.get_logger(__name__)
datasets.logging.set_verbosity_info()


_DESCRIPTION = "XYZ"
_CITATION = "XYZ"
_HOMEPAGE = "https://github.com/ivansipiran/CC6204-Deep-Learning/blob/main/Hackaton/hackaton.md"

_REPO = "https://huggingface.co/datasets/alkzar90/CC6204-Hackaton-Cub-Dataset/resolve/main/data"

_URLS = {
   "train_test_split": f"{_REPO}/train_test_split.txt",
   "classes": f"{_REPO}/classes.txt",
   "image_class_labels": f"{_REPO}/image_class_labels.txt",
   "images": f"{_REPO}/images.txt",
   "image_urls": f"{_REPO}/images.zip",
   "text_urls": f"{_REPO}/text.zip",
}

# Create id-to-label dictionary using the classes file
classes = get(_URLS["classes"]).iter_lines()
logger.info(f"classes: {classes}")

_ID2LABEL = {}
for row in classes:
   row = row.decode("UTF8")
   if row != "":
      idx, label = row.split(" ")
      _ID2LABEL[int(idx)] = re.search("[^\d\.\_+].+", label).group(0).replace("_", " ")
      
logger.info(f"_ID2LABEL: {_ID2LABEL}")

_NAMES = list(_ID2LABEL.values())

# build from images.txt: a mapping from image_file_name -> id
imgpath_to_ids = get(_URLS["images"]).iter_lines()
_IMGNAME2ID = {}
for row in imgpath_to_ids:
   row = row.decode("UTF8")
   if row != "":
      idx, img_name = row.split(" ")
      _IMGNAME2ID[os.path.basename(img_name)] = int(idx)
   


class CubDataset(datasets.GeneratorBasedBuilder):
   """Cub Dataset"""
   
   def _info(self):
      features = datasets.Features({
         "image": datasets.Image(),
         "labels": datasets.features.ClassLabel(names=_NAMES),
      })
      keys = ("image", "labels")
      
      return datasets.DatasetInfo(
         description=_DESCRIPTION,
         features=features,
         supervised_keys=keys,
         homepage=_HOMEPAGE,
         citation=_CITATION,
      )
      
      
   def _split_generators(self, dl_manager):
      # 1: train, 0: test
      train_test_split = get(_URLS["train_test_split"]).iter_lines()
      train_images_idx = set([int(x.decode("UTF8").split(" ")[0]) for x in train_test_split if x.decode("UTF8").split(" ")[1] == "1"])
      logger.info(f"train_images_idx length: {len(train_images_idx)}")
      
      train_files = []
      test_files = []
      
      # Download images
      data_files = dl_manager.download_and_extract(_URLS["image_urls"])
      
      for batch in data_files:
         path_files = dl_manager.iter_files(batch)
         for img in path_files:
            if _IMGNAME2ID[os.path.basename(img)] in train_images_idx:
               train_files.append(img)
            else:
               test_files.append(img)
               
      return [
                 datasets.SplitGenerator(
                    name=datasets.Split.TRAIN,
                    gen_kwargs={
                       "files": train_files
                    }
                 ),
                 datasets.SplitGenerator(
                    name=datasets.Split.TEST,
                    gen_kwargs={
                       "files": test_files
                    }
                 )
      ]
      
      
   def _generate_examples(self, files):
   
      for i, path in enumerate(files):
         file_name = os.path.basename(path)
         if file_name.endswith(".jpg"):
            yield i, {
               "image": path,
               "labels": os.path.basename(os.path.dirname(path)).lower(),
            }