Datasets:
Tasks:
Image-to-Image
Modalities:
Image
Formats:
imagefolder
Size:
10K - 100K
ArXiv:
Tags:
super-resolution
image-quality-assessment
no-reference-image-quality-assessment
mean-opinion-score
License:
Update readme for links addition
Browse files
README.md
CHANGED
|
@@ -14,9 +14,9 @@ tags:
|
|
| 14 |
|
| 15 |
DSR Benchmark is a dataset for studying the quality of images produced by
|
| 16 |
classic and diffusion-based Super-Resolution (SR) methods. The public release
|
| 17 |
-
contains all selected SR images, along with the
|
| 18 |
-
quality scores for 60% of the SR images in each category. The data
|
| 19 |
-
with the remaining 40% are reserved for a future challenge.
|
| 20 |
|
| 21 |
## Dataset summary
|
| 22 |
|
|
@@ -60,6 +60,10 @@ Example:
|
|
| 60 |
00014_vg_dsr.jpg
|
| 61 |
```
|
| 62 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
## LR generation
|
| 64 |
|
| 65 |
Each HR image was downscaled using bicubic interpolation at two scale factors:
|
|
@@ -166,8 +170,8 @@ obtained by aggregating its individual ratings.
|
|
| 166 |
|
| 167 |
Public annotations cover 60% of the Diffusion SR images and 60% of the Classic
|
| 168 |
SR images. For the remaining 40%, the SR images themselves are available, but
|
| 169 |
-
their corresponding MOS values, individual votes, LR images, and HR
|
| 170 |
-
not published.
|
| 171 |
|
| 172 |
> **Note on score representation.** The CSV files contain numeric values in
|
| 173 |
> the form in which they were exported and used to calculate MOS. The data
|
|
@@ -185,9 +189,6 @@ with source ID `00014` are stored in shard `000`. The shard name is denoted by
|
|
| 185 |
.
|
| 186 |
├── README.md
|
| 187 |
├── images/
|
| 188 |
-
│ ├── hr/
|
| 189 |
-
│ │ └── {prefix}/
|
| 190 |
-
│ │ └── xxxxx_vg_dsr.jpg
|
| 191 |
│ ├── lr/
|
| 192 |
│ │ ├── x2/
|
| 193 |
│ │ │ └── {prefix}/
|
|
@@ -206,17 +207,29 @@ with source ID `00014` are stored in shard `000`. The shard name is denoted by
|
|
| 206 |
├── mos.csv
|
| 207 |
├── votes.csv
|
| 208 |
├── mos_csr.csv
|
| 209 |
-
|
|
|
|
| 210 |
```
|
| 211 |
|
| 212 |
The public repository intentionally does not include `split_assignment.csv` or
|
| 213 |
-
`sr_manifest.csv`. Available SR
|
| 214 |
-
five-digit `xxxxx` identifier
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 215 |
|
| 216 |
-
|
| 217 |
|
| 218 |
-
|
| 219 |
-
|
|
|
|
|
|
|
|
|
|
| 220 |
|
| 221 |
### `images/lr/x2/{prefix}/` and `images/lr/x4/{prefix}/`
|
| 222 |
|
|
@@ -286,16 +299,19 @@ the SR image
|
|
| 286 |
CODSR_x2_00014_vg_dsr_blur_x2.jpg
|
| 287 |
```
|
| 288 |
|
| 289 |
-
has the following
|
| 290 |
|
| 291 |
```text
|
| 292 |
images/lr/x2/000/00014_vg_dsr_blur_x2.jpg
|
| 293 |
-
images/hr/000/00014_vg_dsr.jpg
|
| 294 |
```
|
| 295 |
|
| 296 |
-
|
| 297 |
-
|
| 298 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 299 |
|
| 300 |
## Loading the annotations
|
| 301 |
|
|
@@ -314,18 +330,25 @@ classic_mos = pd.read_csv(
|
|
| 314 |
root / "metadata/mos_csr.csv",
|
| 315 |
index_col=0,
|
| 316 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 317 |
|
| 318 |
print(diffusion_mos.head())
|
| 319 |
print(classic_mos.head())
|
|
|
|
| 320 |
```
|
| 321 |
|
| 322 |
-
The
|
|
|
|
| 323 |
|
| 324 |
```python
|
| 325 |
source_id = "00014"
|
| 326 |
prefix = source_id[:3]
|
| 327 |
|
| 328 |
-
|
|
|
|
| 329 |
lr_path = root / "images/lr/x2" / prefix / f"{source_id}_vg_dsr_blur_x2.jpg"
|
| 330 |
sr_path = (
|
| 331 |
root
|
|
@@ -366,8 +389,8 @@ print(source_id) # 00014
|
|
| 366 |
of KonIQ-10k, COCO, AVA, and FLIVE.
|
| 367 |
- The public and challenge partitions are grouped by HR content; therefore,
|
| 368 |
exactly 60% of the HR groups need not be public.
|
| 369 |
-
- Missing LR, HR, or MOS data in the public release are part of the
|
| 370 |
-
protocol and do not indicate dataset corruption.
|
| 371 |
|
| 372 |
## Terms of use and attribution
|
| 373 |
|
|
|
|
| 14 |
|
| 15 |
DSR Benchmark is a dataset for studying the quality of images produced by
|
| 16 |
classic and diffusion-based Super-Resolution (SR) methods. The public release
|
| 17 |
+
contains all selected SR images, along with the LR inputs, HR source links, and
|
| 18 |
+
subjective quality scores for 60% of the SR images in each category. The data
|
| 19 |
+
associated with the remaining 40% are reserved for a future challenge.
|
| 20 |
|
| 21 |
## Dataset summary
|
| 22 |
|
|
|
|
| 60 |
00014_vg_dsr.jpg
|
| 61 |
```
|
| 62 |
|
| 63 |
+
The HR image files are not redistributed in the public repository. For the
|
| 64 |
+
public partition, links to the original HR images or their source pages are
|
| 65 |
+
provided in `metadata/hr_source_links.csv`.
|
| 66 |
+
|
| 67 |
## LR generation
|
| 68 |
|
| 69 |
Each HR image was downscaled using bicubic interpolation at two scale factors:
|
|
|
|
| 170 |
|
| 171 |
Public annotations cover 60% of the Diffusion SR images and 60% of the Classic
|
| 172 |
SR images. For the remaining 40%, the SR images themselves are available, but
|
| 173 |
+
their corresponding MOS values, individual votes, LR images, and HR source
|
| 174 |
+
links are not published.
|
| 175 |
|
| 176 |
> **Note on score representation.** The CSV files contain numeric values in
|
| 177 |
> the form in which they were exported and used to calculate MOS. The data
|
|
|
|
| 189 |
.
|
| 190 |
├── README.md
|
| 191 |
├── images/
|
|
|
|
|
|
|
|
|
|
| 192 |
│ ├── lr/
|
| 193 |
│ │ ├── x2/
|
| 194 |
│ │ │ └── {prefix}/
|
|
|
|
| 207 |
├── mos.csv
|
| 208 |
├── votes.csv
|
| 209 |
├── mos_csr.csv
|
| 210 |
+
├── votes_csr.csv
|
| 211 |
+
└── hr_source_links.csv
|
| 212 |
```
|
| 213 |
|
| 214 |
The public repository intentionally does not include `split_assignment.csv` or
|
| 215 |
+
`sr_manifest.csv`. Available SR and LR images, as well as the corresponding HR
|
| 216 |
+
source links, can be matched using the five-digit `xxxxx` identifier.
|
| 217 |
+
|
| 218 |
+
### `metadata/hr_source_links.csv`
|
| 219 |
+
|
| 220 |
+
Contains source information and links for the HR references associated with
|
| 221 |
+
the public partition. The HR files themselves are not redistributed in this
|
| 222 |
+
repository. Match a row to an LR or SR image using `source_id` or
|
| 223 |
+
`hr_file_name`, then use `source_url` to access the original image or its
|
| 224 |
+
source page.
|
| 225 |
|
| 226 |
+
The `source_url_type` column indicates the type of destination:
|
| 227 |
|
| 228 |
+
- `exact_image` — a direct link to the original image;
|
| 229 |
+
- `exact_image_page` — a page containing the original image;
|
| 230 |
+
- `dataset_page_only` — the source dataset page, when an image-specific link
|
| 231 |
+
is unavailable;
|
| 232 |
+
- `unresolved` — no suitable public link was resolved.
|
| 233 |
|
| 234 |
### `images/lr/x2/{prefix}/` and `images/lr/x4/{prefix}/`
|
| 235 |
|
|
|
|
| 299 |
CODSR_x2_00014_vg_dsr_blur_x2.jpg
|
| 300 |
```
|
| 301 |
|
| 302 |
+
has the following public LR reference:
|
| 303 |
|
| 304 |
```text
|
| 305 |
images/lr/x2/000/00014_vg_dsr_blur_x2.jpg
|
|
|
|
| 306 |
```
|
| 307 |
|
| 308 |
+
The corresponding HR filename is `00014_vg_dsr.jpg`. Its source link can be
|
| 309 |
+
obtained from the row with `source_id=00014` in
|
| 310 |
+
`metadata/hr_source_links.csv`.
|
| 311 |
+
|
| 312 |
+
If the corresponding LR file and HR-link row are absent from the public
|
| 313 |
+
repository, the SR image belongs to the challenge partition, and its MOS and
|
| 314 |
+
votes are not published either.
|
| 315 |
|
| 316 |
## Loading the annotations
|
| 317 |
|
|
|
|
| 330 |
root / "metadata/mos_csr.csv",
|
| 331 |
index_col=0,
|
| 332 |
)
|
| 333 |
+
hr_links = pd.read_csv(
|
| 334 |
+
root / "metadata/hr_source_links.csv",
|
| 335 |
+
dtype={"source_id": str},
|
| 336 |
+
)
|
| 337 |
|
| 338 |
print(diffusion_mos.head())
|
| 339 |
print(classic_mos.head())
|
| 340 |
+
print(hr_links.head())
|
| 341 |
```
|
| 342 |
|
| 343 |
+
The LR and SR paths and the HR source URL can be obtained from the source
|
| 344 |
+
identifier:
|
| 345 |
|
| 346 |
```python
|
| 347 |
source_id = "00014"
|
| 348 |
prefix = source_id[:3]
|
| 349 |
|
| 350 |
+
hr_row = hr_links.loc[hr_links["source_id"] == source_id].iloc[0]
|
| 351 |
+
hr_url = hr_row["source_url"]
|
| 352 |
lr_path = root / "images/lr/x2" / prefix / f"{source_id}_vg_dsr_blur_x2.jpg"
|
| 353 |
sr_path = (
|
| 354 |
root
|
|
|
|
| 389 |
of KonIQ-10k, COCO, AVA, and FLIVE.
|
| 390 |
- The public and challenge partitions are grouped by HR content; therefore,
|
| 391 |
exactly 60% of the HR groups need not be public.
|
| 392 |
+
- Missing LR, HR source-link, or MOS data in the public release are part of the
|
| 393 |
+
challenge protocol and do not indicate dataset corruption.
|
| 394 |
|
| 395 |
## Terms of use and attribution
|
| 396 |
|