robinhad commited on
Commit
b61768d
·
verified ·
1 Parent(s): b8a4157

Upload 5 files

Browse files
.gitattributes CHANGED
@@ -56,3 +56,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
56
  # Video files - compressed
57
  *.mp4 filter=lfs diff=lfs merge=lfs -text
58
  *.webm filter=lfs diff=lfs merge=lfs -text
 
 
56
  # Video files - compressed
57
  *.mp4 filter=lfs diff=lfs merge=lfs -text
58
  *.webm filter=lfs diff=lfs merge=lfs -text
59
+ flickr_annotations_30k_bak.csv filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,18 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Flickr30k
2
+
3
+ Original paper: [From image descriptions to visual denotations: New similarity metrics for semantic inference over event descriptions](https://aclanthology.org/Q14-1006)
4
+
5
+ Homepage: https://shannon.cs.illinois.edu/DenotationGraph/
6
+
7
+ Bibtex:
8
+ ```
9
+ @article{young2014image,
10
+ title={From image descriptions to visual denotations: New similarity metrics for semantic inference over event descriptions},
11
+ author={Young, Peter and Lai, Alice and Hodosh, Micah and Hockenmaier, Julia},
12
+ journal={Transactions of the Association for Computational Linguistics},
13
+ volume={2},
14
+ pages={67--78},
15
+ year={2014},
16
+ publisher={MIT Press}
17
+ }
18
+ ```
flickr30k-images.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2ce2420c0d17f0531deaa89ac657b4d5067ec519da16ff1ea12acbf5619c7391
3
+ size 4390240817
flickr30k.py ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # coding=utf-8
2
+ # Copyright 2022 the HuggingFace Datasets Authors.
3
+ #
4
+ # Licensed under the Apache License, Version 2.0 (the "License");
5
+ # you may not use this file except in compliance with the License.
6
+ # You may obtain a copy of the License at
7
+ #
8
+ # http://www.apache.org/licenses/LICENSE-2.0
9
+ #
10
+ # Unless required by applicable law or agreed to in writing, software
11
+ # distributed under the License is distributed on an "AS IS" BASIS,
12
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
+ # See the License for the specific language governing permissions and
14
+ # limitations under the License.
15
+
16
+ import os
17
+ import pandas as pd
18
+ import datasets
19
+ import json
20
+ import ast
21
+ from huggingface_hub import hf_hub_url
22
+
23
+ _INPUT_CSV = "flickr_annotations_30k.csv"
24
+ _INPUT_IMAGES = "flickr30k-images"
25
+ _REPO_ID = "nlphuji/flickr30k"
26
+ _JSON_KEYS = ["raw", "sentids"]
27
+
28
+
29
+ class Dataset(datasets.GeneratorBasedBuilder):
30
+ VERSION = datasets.Version("1.1.0")
31
+ BUILDER_CONFIGS = [
32
+ datasets.BuilderConfig(name="TEST", version=VERSION, description="test"),
33
+ ]
34
+
35
+ def _info(self):
36
+ return datasets.DatasetInfo(
37
+ features=datasets.Features(
38
+ {
39
+ "image": datasets.Image(),
40
+ "caption": [datasets.Value("string")],
41
+ "sentids": [datasets.Value("string")],
42
+ "split": datasets.Value("string"),
43
+ "img_id": datasets.Value("string"),
44
+ "filename": datasets.Value("string"),
45
+ }
46
+ ),
47
+ task_templates=[],
48
+ )
49
+
50
+ def _split_generators(self, dl_manager):
51
+ """Returns SplitGenerators."""
52
+
53
+ repo_id = _REPO_ID
54
+ data_dir = dl_manager.download_and_extract(
55
+ {
56
+ "examples_csv": hf_hub_url(
57
+ repo_id=repo_id, repo_type="dataset", filename=_INPUT_CSV
58
+ ),
59
+ "images_dir": hf_hub_url(
60
+ repo_id=repo_id,
61
+ repo_type="dataset",
62
+ filename=f"{_INPUT_IMAGES}.zip",
63
+ ),
64
+ }
65
+ )
66
+
67
+ return [datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs=data_dir)]
68
+
69
+ def _generate_examples(self, examples_csv, images_dir):
70
+ """Yields examples."""
71
+ examples_csv = "/home/robinhad/Projects/flickr30k/flickr_annotations_30k.csv"
72
+ df = pd.read_csv(examples_csv)
73
+ for c in _JSON_KEYS:
74
+ df[c] = df[c].apply(ast.literal_eval)
75
+
76
+ for r_idx, r in df.iterrows():
77
+ r_dict = r.to_dict()
78
+ image_path = os.path.join(images_dir, _INPUT_IMAGES, r_dict["filename"])
79
+ r_dict["image"] = image_path
80
+ r_dict["caption"] = r_dict.pop("raw")
81
+ yield r_idx, r_dict
flickr_annotations_30k.csv ADDED
The diff for this file is too large to render. See raw diff
 
flickr_annotations_30k_bak.csv ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:395990db603ab8bafd5c7ab2746b22058bb1e75b78b3eb56ad755931364ac137
3
+ size 12896305