# Pytorch Implementation of Common Image Generation Metrics
## Installation
```
pip install pytorch-image-generation-metrics
```
## Quick Start
```python
from pytorch_image_generation_metrics import get_inception_score, get_fid
images = ... # [N, 3, H, W] normalized to [0, 1]
IS, IS_std = get_inception_score(images) # Inception Score
FID = get_fid(images, 'path/to/fid_ref.npz') # Frechet Inception Distance
```
The file `path/to/fid_ref.npz` is compatiable with the [official FID implementation](https://github.com/bioinf-jku/TTUR).
## Notes
The FID implementation is inspired by [pytorch-fid](https://github.com/mseitzer/pytorch-fid).
This repository is developed for personal research. If you find this package useful, please feel free to open issues.
## Features
- Currently, this package supports the following metrics:
- [Inception Score](https://github.com/openai/improved-gan) (IS)
- [Fréchet Inception Distance](https://github.com/bioinf-jku/TTUR) (FID)
- The computation procedures for IS and FID are integrated to avoid multiple forward passes.
- Supports reading images on the fly to prevent out-of-memory issues, especially for large-scale images.
- Supports computation on GPU to speed up some CPU operations, such as `np.cov` and `scipy.linalg.sqrtm`.
## Reproducing Results of Official Implementations on CIFAR-10
| |Train IS |Test IS |Train(50k) vs Test(10k)<br>FID|
|---------------------|:--------:|:--------:|:----------------------------:|
|Official |11.24±0.20|10.98±0.22|3.1508 |
|ours |11.26±0.13|10.97±0.19|3.1525 |
|ours `use_torch=True`|11.26±0.15|10.97±0.20|3.1457 |
The results differ slightly from the official implementations due to the framework differences between PyTorch and TensorFlow.
## Documentation
### Prepare Statistical Reference for FID
- [Download](https://drive.google.com/drive/folders/1UBdzl6GtNMwNQ5U-4ESlIer43tNjiGJC?usp=sharing) the pre-calculated reference, or
- Calculate the statistical reference for your custom dataset using the command-line tool:
```bash
python -m pytorch_image_generation_metrics.fid_ref \
--path path/to/images \
--output path/to/fid_ref.npz
```
See [fid_ref.py](./pytorch_image_generation_metrics/fid_ref.py) for details.
### Inception Features
- When getting IS or FID, the `InceptionV3` model will be loaded into `torch.device('cuda:0')` by default.
- Change the `device` argument in the `get_*` functions to set the torch device.
### Using `torch.Tensor` as images
- Prepare images as `torch.float32` tensors with shape `[N, 3, H, W]`, normalized to `[0,1]`.
```python
from pytorch_image_generation_metrics import (
get_inception_score,
get_fid,
get_inception_score_and_fid
)
images = ... # [N, 3, H, W]
assert 0 <= images.min() and images.max() <= 1
# Inception Score
IS, IS_std = get_inception_score(
images)
# Frechet Inception Distance
FID = get_fid(
images, 'path/to/fid_ref.npz')
# Inception Score & Frechet Inception Distance
(IS, IS_std), FID = get_inception_score_and_fid(
images, 'path/to/fid_ref.npz')
```
### Using PyTorch DataLoader to Provide Images
1. Use `pytorch_image_generation_metrics.ImageDataset` to collect images from your storage or use your custom `torch.utils.data.Dataset`.
```python
from pytorch_image_generation_metrics import ImageDataset
from torch.utils.data import DataLoader
dataset = ImageDataset(path_to_dir, exts=['png', 'jpg'])
loader = DataLoader(dataset, batch_size=50, num_workers=4)
```
You can wrap a generative model in a dataset to support generating images on the fly.
```python
class GeneratorDataset(Dataset):
def __init__(self, G, noise_dim):
self.G = G
self.noise_dim = noise_dim
def __len__(self):
return 50000
def __getitem__(self, index):
return self.G(torch.randn(1, self.noise_dim))
dataset = GeneratorDataset(G, noise_dim=128)
loader = DataLoader(dataset, batch_size=50, num_workers=0)
```
2. Calculate metrics
```python
from pytorch_image_generation_metrics import (
get_inception_score,
get_fid,
get_inception_score_and_fid
)
# Inception Score
IS, IS_std = get_inception_score(
loader)
# Frechet Inception Distance
FID = get_fid(
loader, 'path/to/fid_ref.npz')
# Inception Score & Frechet Inception Distance
(IS, IS_std), FID = get_inception_score_and_fid(
loader, 'path/to/fid_ref.npz')
```
### Load Images from a Directory
- Calculate metrics for images in a directory and its subfolders.
```python
from pytorch_image_generation_metrics import (
get_inception_score_from_directory,
get_fid_from_directory,
get_inception_score_and_fid_from_directory)
IS, IS_std = get_inception_score_from_directory(
'path/to/images')
FID = get_fid_from_directory(
'path/to/images', 'path/to/fid_ref.npz')
(IS, IS_std), FID = get_inception_score_and_fid_from_directory(
'path/to/images', 'path/to/fid_ref.npz')
```
### Accelerating Matrix Computation with PyTorch
- Set `use_torch=True` when calling functions like `get_inception_score`, `get_fid`, etc.
- **WARNING**: when `use_torch=True` is used, the FID might be `nan` due to the unstable implementation of matrix sqrt root.
## Tested Versions
- `python 3.9 + torch 1.13.1 + CUDA 11.7`
- `python 3.9 + torch 2.3.0 + CUDA 12.1`
## License
This implementation is licensed under the Apache License 2.0.
This implementation is derived from [pytorch-fid](https://github.com/mseitzer/pytorch-fid), licensed under the Apache License 2.0.
FID was introduced by Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler and Sepp Hochreiter in "GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium", see [https://arxiv.org/abs/1706.08500](https://arxiv.org/abs/1706.08500)
The original implementation of FID is by the Institute of Bioinformatics, JKU Linz, licensed under the Apache License 2.0.
See [https://github.com/bioinf-jku/TTUR](https://github.com/bioinf-jku/TTUR).
Raw data
{
"_id": null,
"home_page": "https://github.com/w86763777/pytorch_image_generation_metrics",
"name": "pytorch-image-generation-metrics",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.6",
"maintainer_email": null,
"keywords": "PyTorch, Inception Score, IS, Frechet Inception Distance, FID",
"author": "Yi-Lun Wu",
"author_email": "w86763777@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/d4/ee/ed446ad931b6fb0178739ef79524493247426fa3a40dccc4524ec411aa10/pytorch_image_generation_metrics-0.6.1.tar.gz",
"platform": null,
"description": "# Pytorch Implementation of Common Image Generation Metrics\n\n\n\n## Installation\n```\npip install pytorch-image-generation-metrics\n```\n\n## Quick Start\n```python\nfrom pytorch_image_generation_metrics import get_inception_score, get_fid\n\nimages = ... # [N, 3, H, W] normalized to [0, 1]\nIS, IS_std = get_inception_score(images) # Inception Score\nFID = get_fid(images, 'path/to/fid_ref.npz') # Frechet Inception Distance\n```\nThe file `path/to/fid_ref.npz` is compatiable with the [official FID implementation](https://github.com/bioinf-jku/TTUR).\n\n## Notes\nThe FID implementation is inspired by [pytorch-fid](https://github.com/mseitzer/pytorch-fid).\n\nThis repository is developed for personal research. If you find this package useful, please feel free to open issues.\n\n## Features\n- Currently, this package supports the following metrics:\n - [Inception Score](https://github.com/openai/improved-gan) (IS)\n - [Fr\u00e9chet Inception Distance](https://github.com/bioinf-jku/TTUR) (FID)\n- The computation procedures for IS and FID are integrated to avoid multiple forward passes.\n- Supports reading images on the fly to prevent out-of-memory issues, especially for large-scale images.\n- Supports computation on GPU to speed up some CPU operations, such as `np.cov` and `scipy.linalg.sqrtm`.\n\n## Reproducing Results of Official Implementations on CIFAR-10\n\n| |Train IS |Test IS |Train(50k) vs Test(10k)<br>FID|\n|---------------------|:--------:|:--------:|:----------------------------:|\n|Official |11.24\u00b10.20|10.98\u00b10.22|3.1508 |\n|ours |11.26\u00b10.13|10.97\u00b10.19|3.1525 |\n|ours `use_torch=True`|11.26\u00b10.15|10.97\u00b10.20|3.1457 |\n\nThe results differ slightly from the official implementations due to the framework differences between PyTorch and TensorFlow.\n\n## Documentation\n\n### Prepare Statistical Reference for FID\n- [Download](https://drive.google.com/drive/folders/1UBdzl6GtNMwNQ5U-4ESlIer43tNjiGJC?usp=sharing) the pre-calculated reference, or\n- Calculate the statistical reference for your custom dataset using the command-line tool:\n ```bash\n python -m pytorch_image_generation_metrics.fid_ref \\\n --path path/to/images \\\n --output path/to/fid_ref.npz\n ```\n See [fid_ref.py](./pytorch_image_generation_metrics/fid_ref.py) for details.\n\n### Inception Features\n- When getting IS or FID, the `InceptionV3` model will be loaded into `torch.device('cuda:0')` by default.\n- Change the `device` argument in the `get_*` functions to set the torch device.\n\n### Using `torch.Tensor` as images\n\n- Prepare images as `torch.float32` tensors with shape `[N, 3, H, W]`, normalized to `[0,1]`.\n ```python\n from pytorch_image_generation_metrics import (\n get_inception_score,\n get_fid,\n get_inception_score_and_fid\n )\n\n images = ... # [N, 3, H, W]\n assert 0 <= images.min() and images.max() <= 1\n\n # Inception Score\n IS, IS_std = get_inception_score(\n images)\n\n # Frechet Inception Distance\n FID = get_fid(\n images, 'path/to/fid_ref.npz')\n\n # Inception Score & Frechet Inception Distance\n (IS, IS_std), FID = get_inception_score_and_fid(\n images, 'path/to/fid_ref.npz')\n\n ```\n\n### Using PyTorch DataLoader to Provide Images\n\n1. Use `pytorch_image_generation_metrics.ImageDataset` to collect images from your storage or use your custom `torch.utils.data.Dataset`.\n ```python\n from pytorch_image_generation_metrics import ImageDataset\n from torch.utils.data import DataLoader\n\n dataset = ImageDataset(path_to_dir, exts=['png', 'jpg'])\n loader = DataLoader(dataset, batch_size=50, num_workers=4)\n ```\n\n You can wrap a generative model in a dataset to support generating images on the fly.\n ```python\n class GeneratorDataset(Dataset):\n def __init__(self, G, noise_dim):\n self.G = G\n self.noise_dim = noise_dim\n\n def __len__(self):\n return 50000\n\n def __getitem__(self, index):\n return self.G(torch.randn(1, self.noise_dim))\n\n dataset = GeneratorDataset(G, noise_dim=128)\n loader = DataLoader(dataset, batch_size=50, num_workers=0)\n ```\n\n2. Calculate metrics\n ```python\n from pytorch_image_generation_metrics import (\n get_inception_score,\n get_fid,\n get_inception_score_and_fid\n )\n\n # Inception Score\n IS, IS_std = get_inception_score(\n loader)\n\n # Frechet Inception Distance\n FID = get_fid(\n loader, 'path/to/fid_ref.npz')\n\n # Inception Score & Frechet Inception Distance\n (IS, IS_std), FID = get_inception_score_and_fid(\n loader, 'path/to/fid_ref.npz')\n ```\n\n### Load Images from a Directory\n\n- Calculate metrics for images in a directory and its subfolders.\n ```python\n from pytorch_image_generation_metrics import (\n get_inception_score_from_directory,\n get_fid_from_directory,\n get_inception_score_and_fid_from_directory)\n\n IS, IS_std = get_inception_score_from_directory(\n 'path/to/images')\n FID = get_fid_from_directory(\n 'path/to/images', 'path/to/fid_ref.npz')\n (IS, IS_std), FID = get_inception_score_and_fid_from_directory(\n 'path/to/images', 'path/to/fid_ref.npz')\n ```\n\n### Accelerating Matrix Computation with PyTorch\n\n- Set `use_torch=True` when calling functions like `get_inception_score`, `get_fid`, etc.\n\n- **WARNING**: when `use_torch=True` is used, the FID might be `nan` due to the unstable implementation of matrix sqrt root.\n\n## Tested Versions\n- `python 3.9 + torch 1.13.1 + CUDA 11.7`\n- `python 3.9 + torch 2.3.0 + CUDA 12.1`\n\n## License\n\nThis implementation is licensed under the Apache License 2.0.\n\nThis implementation is derived from [pytorch-fid](https://github.com/mseitzer/pytorch-fid), licensed under the Apache License 2.0.\n\nFID was introduced by Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler and Sepp Hochreiter in \"GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium\", see [https://arxiv.org/abs/1706.08500](https://arxiv.org/abs/1706.08500)\n\nThe original implementation of FID is by the Institute of Bioinformatics, JKU Linz, licensed under the Apache License 2.0.\nSee [https://github.com/bioinf-jku/TTUR](https://github.com/bioinf-jku/TTUR).\n\n\n",
"bugtrack_url": null,
"license": null,
"summary": "Package for calculating image generation metrics using Pytorch",
"version": "0.6.1",
"project_urls": {
"Homepage": "https://github.com/w86763777/pytorch_image_generation_metrics"
},
"split_keywords": [
"pytorch",
" inception score",
" is",
" frechet inception distance",
" fid"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "df379b1376b1b2bebcdb3b5a45ff4a70313fa2cc8b22caeb217ed4b3c4d55116",
"md5": "2a5413d2aaf93281f21fbe4bff5ee40c",
"sha256": "6635307f4878866b25c835baf919253f3baea3e7bc55153f5e7e1e2a697d0073"
},
"downloads": -1,
"filename": "pytorch_image_generation_metrics-0.6.1-py3-none-any.whl",
"has_sig": false,
"md5_digest": "2a5413d2aaf93281f21fbe4bff5ee40c",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.6",
"size": 25320,
"upload_time": "2024-06-14T03:20:08",
"upload_time_iso_8601": "2024-06-14T03:20:08.175401Z",
"url": "https://files.pythonhosted.org/packages/df/37/9b1376b1b2bebcdb3b5a45ff4a70313fa2cc8b22caeb217ed4b3c4d55116/pytorch_image_generation_metrics-0.6.1-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "d4eeed446ad931b6fb0178739ef79524493247426fa3a40dccc4524ec411aa10",
"md5": "0e05d74ec1d08d6fd30ca9f5e6616507",
"sha256": "2e81f81de0fe331e7c769b203695c2a8c60ace89b6229141c6774a31a49c13c6"
},
"downloads": -1,
"filename": "pytorch_image_generation_metrics-0.6.1.tar.gz",
"has_sig": false,
"md5_digest": "0e05d74ec1d08d6fd30ca9f5e6616507",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.6",
"size": 21849,
"upload_time": "2024-06-14T03:20:10",
"upload_time_iso_8601": "2024-06-14T03:20:10.339506Z",
"url": "https://files.pythonhosted.org/packages/d4/ee/ed446ad931b6fb0178739ef79524493247426fa3a40dccc4524ec411aa10/pytorch_image_generation_metrics-0.6.1.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-06-14 03:20:10",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "w86763777",
"github_project": "pytorch_image_generation_metrics",
"github_not_found": true,
"lcname": "pytorch-image-generation-metrics"
}
JSON
