Update README.md
Browse files
README.md
CHANGED
@@ -7,4 +7,33 @@ language:
|
|
7 |
pretty_name: flickr8k
|
8 |
size_categories:
|
9 |
- 1K<n<10K
|
10 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
pretty_name: flickr8k
|
8 |
size_categories:
|
9 |
- 1K<n<10K
|
10 |
+
---
|
11 |
+
|
12 |
+
You must download the dataset files manually. You can visit [this](https://github.com/jbrownlee/Datasets/releases/tag/Flickr8k) page or run `download.sh` to get files.
|
13 |
+
|
14 |
+
After, you can load dataset by referencing the directory:
|
15 |
+
|
16 |
+
```py
|
17 |
+
import datasets
|
18 |
+
ds = datasets.load_dataset("atasoglu/flickr8k-dataset", data_dir="data")
|
19 |
+
print(ds)
|
20 |
+
```
|
21 |
+
|
22 |
+
```
|
23 |
+
DatasetDict({
|
24 |
+
train: Dataset({
|
25 |
+
features: ['image_id', 'image_path', 'captions'],
|
26 |
+
num_rows: 6000
|
27 |
+
})
|
28 |
+
test: Dataset({
|
29 |
+
features: ['image_id', 'image_path', 'captions'],
|
30 |
+
num_rows: 1000
|
31 |
+
})
|
32 |
+
validation: Dataset({
|
33 |
+
features: ['image_id', 'image_path', 'captions'],
|
34 |
+
num_rows: 1000
|
35 |
+
})
|
36 |
+
})
|
37 |
+
```
|
38 |
+
|
39 |
+
I don't own the copyright of the images. Please [visit](https://forms.illinois.edu/sec/1713398) for more.
|