Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,132 @@
|
|
| 1 |
---
|
| 2 |
license: cc
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: cc
|
| 3 |
---
|
| 4 |
+
|
| 5 |
+
# Google WIT Vietnamese
|
| 6 |
+
|
| 7 |
+
This data repos contain extracted data from [Google WIT](https://github.com/google-research-datasets/wit/blob/main/DATA.md). The extracted data is all for Vietnamese language.
|
| 8 |
+
|
| 9 |
+
Given `x` is a data point in the OG dataset which has keys following OG `field_name`, the criteria to filter is
|
| 10 |
+
```python
|
| 11 |
+
criteria = lambda x: x.get("language", "") == "vi" and x.get("caption_reference_description", "")
|
| 12 |
+
```
|
| 13 |
+
|
| 14 |
+
## Text-related details
|
| 15 |
+
|
| 16 |
+
All `.tsv.gz` files follow OG data files in terms of file names and file structures.
|
| 17 |
+
|
| 18 |
+
### Train split
|
| 19 |
+
|
| 20 |
+
`wit_v1.train.*.tsv.gz`
|
| 21 |
+
|
| 22 |
+
Train data length of each file (not including the header),
|
| 23 |
+
```
|
| 24 |
+
17690
|
| 25 |
+
17756
|
| 26 |
+
17810
|
| 27 |
+
17724
|
| 28 |
+
17619
|
| 29 |
+
17494
|
| 30 |
+
17624
|
| 31 |
+
17696
|
| 32 |
+
17777
|
| 33 |
+
17562
|
| 34 |
+
```
|
| 35 |
+
Total 176752
|
| 36 |
+
|
| 37 |
+
### Validation split
|
| 38 |
+
|
| 39 |
+
`wit_v1.val.*.tsv.gz`
|
| 40 |
+
|
| 41 |
+
Val data length of each file (not including the header),
|
| 42 |
+
```
|
| 43 |
+
292
|
| 44 |
+
273
|
| 45 |
+
275
|
| 46 |
+
320
|
| 47 |
+
306
|
| 48 |
+
```
|
| 49 |
+
Total 1466
|
| 50 |
+
|
| 51 |
+
### Test split
|
| 52 |
+
|
| 53 |
+
`wit_v1.test.*.tsv.gz`
|
| 54 |
+
|
| 55 |
+
Test data length of each file (not including the header),
|
| 56 |
+
```
|
| 57 |
+
215
|
| 58 |
+
202
|
| 59 |
+
201
|
| 60 |
+
201
|
| 61 |
+
229
|
| 62 |
+
```
|
| 63 |
+
Total 1048
|
| 64 |
+
|
| 65 |
+
## Image-related details
|
| 66 |
+
|
| 67 |
+
### Image URL only
|
| 68 |
+
|
| 69 |
+
`*.image_url_list.txt` are simply lists of image urls from `*.tsv.gz` files
|
| 70 |
+
|
| 71 |
+
Image url length of each file (train, val, test, all)
|
| 72 |
+
```
|
| 73 |
+
157281
|
| 74 |
+
1271
|
| 75 |
+
900
|
| 76 |
+
159452
|
| 77 |
+
```
|
| 78 |
+
Google Research has made sure that all sets don't share same exact images.
|
| 79 |
+
|
| 80 |
+
### Downloaded Images
|
| 81 |
+
|
| 82 |
+
⚠ Please for the love of the gods, read this section carefully.
|
| 83 |
+
|
| 84 |
+
For `all.index.fmt_id.image_url_list.tsv`, from left to right, without headers, the columns are `index`, `fmt_id`, `image_url`. It is to map `image_url` (in `all.image_url_list.txt`) to `fmt_id`. It's for downloading images.
|
| 85 |
+
|
| 86 |
+
`fmt_id` is:
|
| 87 |
+
- used to name images (with proper image extensions) in `images/`.
|
| 88 |
+
- `index` but filled with 6 zeros
|
| 89 |
+
|
| 90 |
+
Downloading time was less than 36 hours with:
|
| 91 |
+
- 90 Mbps
|
| 92 |
+
- Processor Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz 1.99 GHz
|
| 93 |
+
- No asynchronous
|
| 94 |
+
|
| 95 |
+
For `fail.index.fmt_id.status.image_url_list.tsv`, from left to right, without headers, the columns are `index`, `fmt_id`, `status`, `image_url`. It is to track image urls (during downloading) that are inaccessible.
|
| 96 |
+
|
| 97 |
+
3367 image urls returned 404 (`status` values). In other words, we were able to download 97.88839275% of images.
|
| 98 |
+
|
| 99 |
+
`images/` folder takes disk space of:
|
| 100 |
+
- 215 GBs (uncompressed)
|
| 101 |
+
- 209 GBs (compressed)
|
| 102 |
+
|
| 103 |
+
We use Pillow to open each image to make sure that downloaded images are usable. We also log all faulty files in `corrupted_image_list.json`. There are less than 70 image files.
|
| 104 |
+
|
| 105 |
+
For `corrupted_image_list.json`, for each item in this list, the keys are `file_name`, `error`. `file_name` is `fmt_id` with extension but without `images/`. Some errors are either:
|
| 106 |
+
- files exceed Pillow default limit
|
| 107 |
+
- files are truncated
|
| 108 |
+
|
| 109 |
+
To actually load those files, the following code can be used to change Pillow behavior
|
| 110 |
+
```python
|
| 111 |
+
from PIL import Image, ImageFile
|
| 112 |
+
|
| 113 |
+
# For very big image files
|
| 114 |
+
Image.MAX_IMAGE_PIXELS = None
|
| 115 |
+
|
| 116 |
+
# For truncated image files
|
| 117 |
+
ImageFile.LOAD_TRUNCATED_IMAGES = True
|
| 118 |
+
```
|
| 119 |
+
|
| 120 |
+
Zip `images/` folder,
|
| 121 |
+
```bash
|
| 122 |
+
zip -r images.zip images/
|
| 123 |
+
zip images.zip --out spanned_images.zip -s 40g
|
| 124 |
+
```
|
| 125 |
+
https://superuser.com/questions/336219/how-do-i-split-a-zip-file-into-multiple-segments
|
| 126 |
+
|
| 127 |
+
Unzip `spanned_images.*` files,
|
| 128 |
+
```bash
|
| 129 |
+
zip -s 0 spanned_images.zip --out images.zip
|
| 130 |
+
unzip images.zip
|
| 131 |
+
```
|
| 132 |
+
https://unix.stackexchange.com/questions/40480/how-to-unzip-a-multipart-spanned-zip-on-linux
|