Kaggle / Kaggle/docker-python

Add imagecodecs packages

Offen
#1,439 1 Kommentar 0 Reaktionen 0 zugewiesene Personen Auf GitHub ansehen
bug help wanted
Vorherrschende Sprache
Python
Sterne
2.7k
Forks
1k
Ø Merge
7 T. 14 Std.
Gemergte PRs (30 T.)
2

Beschreibung

## 🐛 Bug

TLDR error: `imagecodecs` is not found by `tifffile` for decoding input images.

Error trace:
```
/opt/conda/lib/python3.10/site-packages/imageio/v3.py in ?(uri, index, plugin, extension, format_hint, **kwargs)
50 if index is not None:
51 call_kwargs["index"] = index
52
53 with imopen(uri, "r", **plugin_kwargs) as img_file:
---> 54 return np.asarray(img_file.read(**call_kwargs))

/opt/conda/lib/python3.10/site-packages/imageio/plugins/tifffile_v3.py in ?(self, index, page, **kwargs)
165 # read all series in the file and return them as a batch
166 ndimage = np.stack([x for x in self.iter(**kwargs)])
167 else:
168 index = None if index is Ellipsis else index
--> 169 ndimage = self._fh.asarray(series=index, **kwargs)
170
171 return ndimage

/opt/conda/lib/python3.10/site-packages/tifffile/tifffile.py in ?(self, key, series, level, squeeze, out, maxworkers, buffersize)

/opt/conda/lib/python3.10/site-packages/tifffile/tifffile.py in ?(self, out, squeeze, lock, maxworkers, buffersize)

/opt/conda/lib/python3.10/site-packages/tifffile/tifffile.py in ?(self, lock, maxworkers, func, sort, buffersize, _fullsize)

/opt/conda/lib/python3.10/site-packages/tifffile/tifffile.py in ?(args, decodeargs, decode)
9179 def get_resolution(
-> 9180 self,
9181 unit: RESUNIT | int | str | None = None,
9182 scale: float | int | None = None,
9183 ) -> tuple[int | float, int | float]:

/opt/conda/lib/python3.10/site-packages/tifffile/tifffile.py in ?(exc, *args, **kwargs)
8603 def unpack(data: bytes, /) -> NDArray[Any]:
-> 8604 # return NumPy array from packed integers
8605 return imagecodecs.packints_decode(
8606 data, dtype, self.bitspersample, runlen=stwidth * samples
8607 )

ValueError: requires the 'imagecodecs' package
```

### To Reproduce

```python
import imageio.v3 as imageio
!wget -q https://owncloud.gwdg.de/index.php/s/3gAgXzzObADpAoL/download -O data.tif
image = imageio.imread("data.tif")
```

### Expected behavior

This works seamlessly in my local system.

```bash
(sam) anwai@anwai:~/data/test$ python
Python 3.10.0 | packaged by conda-forge | (default, Nov 20 2021, 02:24:10) [GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import imageio.v3 as imageio
>>> import subprocess
>>> subprocess.run(["wget", "https://owncloud.gwdg.de/index.php/s/3gAgXzzObADpAoL/download", "-O", "data.tif"])
--2024-11-13 18:34:14-- https://owncloud.gwdg.de/index.php/s/3gAgXzzObADpAoL/download
Resolving owncloud.gwdg.de (owncloud.gwdg.de)... 134.76.23.45
Connecting to owncloud.gwdg.de (owncloud.gwdg.de)|134.76.23.45|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 215312 (210K) [image/tiff]
Saving to: ‘data.tif’

data.tif 100%[================================================================>] 210,27K --.-KB/s in 0,02s

2024-11-13 18:34:15 (12,5 MB/s) - ‘data.tif’ saved [215312/215312]

CompletedProcess(args=['wget', 'https://owncloud.gwdg.de/index.php/s/3gAgXzzObADpAoL/download', '-O', 'data.tif'], returncode=0)
>>> image = imageio.imread("data.tif")
>>> image.shape
(512, 512)
```

### Additional context

Seems like the package was removed in this PR (https://github.com/Kaggle/docker-python/pull/1425).

Also, installing this using `pip` or `conda` (using `pip install imagecodecs` / `conda install -c conda-forge imagecodecs`) does not seem to resolve the issue.

Thanks in advance!

Beitragsleitfaden

Für dieses Repository ist kein Beitragsleitfaden indexiert

Rechercherichtung

Beginne mit der Überprüfung der Änderung zur Entfernung von Paketen in Kaggle/docker-python PR #1425 und reproduziere den gemeldeten Fehler von imageio.imread("data.tif") mit dem bereitgestellten TIFF-Download. Ermittle, wo dieses Image Python-Pakete bereitstellt, stelle imagecodecs dort wieder her und überprüfe, dass die Reproduktion das TIFF mit shape (512, 512) dekodiert.

Vom Indexierungsmodell aus dem Issue-Text verfasst.

Bewertung

Tech-Stack
docker, python
Bereich
build-system, infrastructure
Issue-Typ
Bug
Schwierigkeit
3/5
Geschätzter Aufwand
1-2 Tage
Aktivitätsstatus
Veraltet
Klarheit
Größtenteils klar
Anfängerfreundlichkeit
35/100

Neue Issues direkt in Ihr Postfach

Eine kurze Übersicht über anfängerfreundliche GitHub-Issues.