# DeepDC: Deep Distance Correlation as a Perceptual Image Quality Evaluator
This is the repository of paper [DeepDC: Deep Distance Correlation as a Perceptual Image Quality Evaluator](https://arxiv.org/abs/***).
### Highlights:
* A novel FR-IQA model that fully utilizes the *texture-sensitiv*e of pre-trained DNN features, which computes **distance correlation** in the deep feature domain
* The model is **exclusively** based on the features of the pre-trained DNNs and does not rely on fine-tuning with MOSs
* Extensive experiments achieve superior performance on five standard IQA datasets, one perceptual similarity dataset, two texture similarity datasets, and one geometric transformation dataset
* It can be employed as an objective function in texture synthesis and neural style transfer
### ====== Pytorch Implementation ======
**Installation:**
- ```pip install DeepDC-PyTorch```
### Requirements:
- Python >= 3.6
- PyTorch >= 1.0
**Usage:**
```python
from DeepDC_PyTorch import DeepDC
model = DeepDC()
# calculate DeepDC between X, Y (a batch of RGB images, data range: 0~1)
deepdc_score = model(X, Y)
```
or
```bash
git clone https://github.com/h4nwei/DeepDC
cd DeepDC_PyTorch
python DeepDC.py --ref <ref_path> --dist <dist_path>
```
## Reference
- R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in *IEEE Conference on Computer Vision and Pattern Recognition*, 2018, pp. 586–595.
- K. Ding, K. Ma, S. Wang, and E. P. Simoncelli, “Image quality assessment: Unifying structure and texture similarity,” *IEEE Transactions on Pattern Analysis and Machine Intelligence*, vol. 44, no. 5, pp. 2567–2581, 2020.
- I. Kligvasser, T. Shaham, Y. Bahat, and T. Michaeli, “Deep selfdissimilarities as powerful visual fingerprints,” in *Neural Information Processing Systems*, 2021, pp. 3939–3951.
## Citation
```bibtex
@article{zhu2023DeepDC,
title={DeepDC: Deep Distance Correlation as a Perceptual Image Quality Evaluator},
author={Zhu, Hanwei and Chen, Baoliang and Zhu, Lingyu and Wang, Shiqi and Lin, Weisi},
journal={CoRR},
volume = {abs/2211.04927v2},
year={2023},
url = {https://arxiv.org/pdf/2211.04927v2.pdf}
}
Raw data
{
"_id": null,
"home_page": "https://github.com/h4nwei/DeepDC",
"name": "DeepDC-PyTorch",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "pytorch,similarity,VGG,texture,structure,metric",
"author": "Hanwei ZHU",
"author_email": "hanwei.zhu@outlook.com",
"download_url": "https://files.pythonhosted.org/packages/a7/1e/0bd328acc4ce7c2ce41c45f9258e933147bd8efdfedcd2cf250d4de61e9b/DeepDC_PyTorch-0.2.tar.gz",
"platform": "python",
"description": "# DeepDC: Deep Distance Correlation as a Perceptual Image Quality Evaluator\n\nThis is the repository of paper [DeepDC: Deep Distance Correlation as a Perceptual Image Quality Evaluator](https://arxiv.org/abs/***). \n\n### Highlights:\n\n* A novel FR-IQA model that fully utilizes the *texture-sensitiv*e of pre-trained DNN features, which computes **distance correlation** in the deep feature domain \n* The model is **exclusively** based on the features of the pre-trained DNNs and does not rely on fine-tuning with MOSs\n* Extensive experiments achieve superior performance on five standard IQA datasets, one perceptual similarity dataset, two texture similarity datasets, and one geometric transformation dataset\n* It can be employed as an objective function in texture synthesis and neural style transfer\n\n\n\n\n### ====== Pytorch Implementation ======\n**Installation:** \n- ```pip install DeepDC-PyTorch```\n\n### Requirements: \n- Python >= 3.6\n- PyTorch >= 1.0\n\n**Usage:** \n```python\nfrom DeepDC_PyTorch import DeepDC\nmodel = DeepDC()\n# calculate DeepDC between X, Y (a batch of RGB images, data range: 0~1) \ndeepdc_score = model(X, Y)\n```\nor\n\n```bash\ngit clone https://github.com/h4nwei/DeepDC\ncd DeepDC_PyTorch\npython DeepDC.py --ref <ref_path> --dist <dist_path>\n```\n\n\n## Reference\n\n- R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, \u201cThe unreasonable effectiveness of deep features as a perceptual metric,\u201d in *IEEE Conference on Computer Vision and Pattern Recognition*, 2018, pp. 586\u2013595.\n- K. Ding, K. Ma, S. Wang, and E. P. Simoncelli, \u201cImage quality assessment: Unifying structure and texture similarity,\u201d *IEEE Transactions on Pattern Analysis and Machine Intelligence*, vol. 44, no. 5, pp. 2567\u20132581, 2020.\n- I. Kligvasser, T. Shaham, Y. Bahat, and T. Michaeli, \u201cDeep selfdissimilarities as powerful visual fingerprints,\u201d in *Neural Information Processing Systems*, 2021, pp. 3939\u20133951.\n\n## Citation\n```bibtex\n@article{zhu2023DeepDC,\ntitle={DeepDC: Deep Distance Correlation as a Perceptual Image Quality Evaluator},\nauthor={Zhu, Hanwei and Chen, Baoliang and Zhu, Lingyu and Wang, Shiqi and Lin, Weisi},\njournal={CoRR},\nvolume = {abs/2211.04927v2},\nyear={2023},\nurl = {https://arxiv.org/pdf/2211.04927v2.pdf}\n}\n\n\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "DeepDC: Deep Distance Correlation as a Perceptual Image Quality Evaluator",
"version": "0.2",
"project_urls": {
"Homepage": "https://github.com/h4nwei/DeepDC"
},
"split_keywords": [
"pytorch",
"similarity",
"vgg",
"texture",
"structure",
"metric"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "add3b476a4f4d9cd43204e9511c917a77de3c4921c84952a74c84185e26d42a0",
"md5": "937e03940e865b2c0c82936b88c81238",
"sha256": "1925bf9ac6dee11b9127b45e5009830b268b9b4f5328e0c7d720fa658cb92819"
},
"downloads": -1,
"filename": "DeepDC_PyTorch-0.2-py3-none-any.whl",
"has_sig": false,
"md5_digest": "937e03940e865b2c0c82936b88c81238",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 5618,
"upload_time": "2023-11-24T13:32:00",
"upload_time_iso_8601": "2023-11-24T13:32:00.576666Z",
"url": "https://files.pythonhosted.org/packages/ad/d3/b476a4f4d9cd43204e9511c917a77de3c4921c84952a74c84185e26d42a0/DeepDC_PyTorch-0.2-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "a71e0bd328acc4ce7c2ce41c45f9258e933147bd8efdfedcd2cf250d4de61e9b",
"md5": "f5702410df7916ce8beec4829a26ad00",
"sha256": "162faa3e0cee26cf4f0c7c89211dc2ded6605b7c02187d25a08f4e5ce82426d7"
},
"downloads": -1,
"filename": "DeepDC_PyTorch-0.2.tar.gz",
"has_sig": false,
"md5_digest": "f5702410df7916ce8beec4829a26ad00",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 4522,
"upload_time": "2023-11-24T13:32:01",
"upload_time_iso_8601": "2023-11-24T13:32:01.811755Z",
"url": "https://files.pythonhosted.org/packages/a7/1e/0bd328acc4ce7c2ce41c45f9258e933147bd8efdfedcd2cf250d4de61e9b/DeepDC_PyTorch-0.2.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-11-24 13:32:01",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "h4nwei",
"github_project": "DeepDC",
"github_not_found": true,
"lcname": "deepdc-pytorch"
}
JSON
