# DOLG in torch and tensorflow (TF2)
Re-implementation (Non Official) of the paper DOLG: Single-Stage Image Retrieval with Deep Orthogonal Fusion of Local and Global Features accepted at ICCV 2021.
[paper](https://arxiv.org/pdf/2108.02927.pdf)
The pytorch checkpoint has been converted into tensorflow format (.h5) from this repository : https://github.com/feymanpriv/DOLG (Official)
## Pipeline
## Installation
> pip install opencv-python==4.5.5.64
> pip install huggingface-hub
to install dolg :
> pip install dolg
OR
> pip install -e .
## Inference
To do some inference on single sample, you can use python script in examples/ folder or use as follows:
```
import dolg
import numpy as np
from dolg.utils.extraction import process_data
depth = 50
# for pytorch
import torch
from dolg.dolg_model_pt import DOLG
from dolg.resnet_pt import ResNet
backbone = ResNet(depth=depth, num_groups=1, width_per_group=64, bn_eps=1e-5,
bn_mom=0.1, trans_fun="bottleneck_transform")
model = DOLG(backbone, s4_dim=2048, s3_dim=1024, s2_dim=512, head_reduction_dim=512,
with_ma=False, num_classes=None, pretrained=f"r{depth}")
img = process_data("image.jpg", "", mode="pt").unsqueeze(0)
with torch.no_grad():
output = model(img)
print(output)
# for tensorflow
import tensorflow as tf
from dolg.dolg_model_tf2 import DOLG
from dolg.resnet_tf2 import ResNet
backbone = ResNet(depth=depth, num_groups=1, width_per_group=64, bn_eps=1e-5,
bn_mom=0.1, trans_fun="bottleneck_transform", name="globalmodel")
model = DOLG(backbone, s4_dim=2048, s3_dim=1024, s2_dim=512, head_reduction_dim=512,
with_ma=False, num_classes=None, pretrained=f"r{depth}")
img = process_data("image.jpg", "", mode="tf")
img = np.expand_dims(img, axis=0)
output = model.predict(img)
print(output)
```
## Data
The model has been trained on google landmark v2. You can find the dataset on the official repository : https://github.com/cvdfoundation/google-landmark .
# Citation :
```bibtex
@misc{yang2021dolg,
title={DOLG: Single-Stage Image Retrieval with Deep Orthogonal Fusion of Local and Global Features},
author={Min Yang and Dongliang He and Miao Fan and Baorong Shi and Xuetong Xue and Fu Li and Errui Ding and Jizhou Huang},
year={2021},
eprint={2108.02927},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
@misc{https://doi.org/10.48550/arxiv.2004.01804,
doi = {10.48550/ARXIV.2004.01804},
url = {https://arxiv.org/abs/2004.01804},
author = {Weyand, Tobias and Araujo, Andre and Cao, Bingyi and Sim, Jack},
keywords = {Computer Vision and Pattern Recognition (cs.CV), FOS: Computer and information sciences, FOS: Computer and information sciences},
title = {Google Landmarks Dataset v2 -- A Large-Scale Benchmark for Instance-Level Recognition and Retrieval},
```
Raw data
{
"_id": null,
"home_page": "https://github.com/Shiro-LK/python-DOLG",
"name": "dolg",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "DOLG for torch and tensorflow,pretrained weights,tensorflow,tf,pytorch,torch",
"author": "Shiro-LK",
"author_email": "shirosaki94@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/2a/80/7ee6e1fa67130ac85949e1eab15e9bc4ebeb720c8f3d504f2fad97a3161e/dolg-0.1.5.1.tar.gz",
"platform": null,
"description": "# DOLG in torch and tensorflow (TF2)\n\nRe-implementation (Non Official) of the paper DOLG: Single-Stage Image Retrieval with Deep Orthogonal Fusion of Local and Global Features accepted at ICCV 2021.\n[paper](https://arxiv.org/pdf/2108.02927.pdf)\n\nThe pytorch checkpoint has been converted into tensorflow format (.h5) from this repository : https://github.com/feymanpriv/DOLG (Official) \n\n## Pipeline\n\n\n\n## Installation \n\n> pip install opencv-python==4.5.5.64\n\n> pip install huggingface-hub\n\nto install dolg : \n\n> pip install dolg\nOR \n> pip install -e .\n\n## Inference\n\nTo do some inference on single sample, you can use python script in examples/ folder or use as follows:\n\n```\nimport dolg\nimport numpy as np\nfrom dolg.utils.extraction import process_data\n\ndepth = 50\n\n# for pytorch\n\nimport torch\nfrom dolg.dolg_model_pt import DOLG\nfrom dolg.resnet_pt import ResNet\n\nbackbone = ResNet(depth=depth, num_groups=1, width_per_group=64, bn_eps=1e-5, \n bn_mom=0.1, trans_fun=\"bottleneck_transform\")\nmodel = DOLG(backbone, s4_dim=2048, s3_dim=1024, s2_dim=512, head_reduction_dim=512,\n with_ma=False, num_classes=None, pretrained=f\"r{depth}\")\nimg = process_data(\"image.jpg\", \"\", mode=\"pt\").unsqueeze(0)\n\nwith torch.no_grad():\n output = model(img)\nprint(output)\n\n# for tensorflow\n\nimport tensorflow as tf\nfrom dolg.dolg_model_tf2 import DOLG\nfrom dolg.resnet_tf2 import ResNet\n\n\nbackbone = ResNet(depth=depth, num_groups=1, width_per_group=64, bn_eps=1e-5, \n bn_mom=0.1, trans_fun=\"bottleneck_transform\", name=\"globalmodel\")\nmodel = DOLG(backbone, s4_dim=2048, s3_dim=1024, s2_dim=512, head_reduction_dim=512,\n with_ma=False, num_classes=None, pretrained=f\"r{depth}\")\nimg = process_data(\"image.jpg\", \"\", mode=\"tf\")\nimg = np.expand_dims(img, axis=0)\noutput = model.predict(img)\nprint(output)\n```\n\n## Data \n\nThe model has been trained on google landmark v2. You can find the dataset on the official repository : https://github.com/cvdfoundation/google-landmark .\n\n\n# Citation : \n\n```bibtex\n\n@misc{yang2021dolg,\n title={DOLG: Single-Stage Image Retrieval with Deep Orthogonal Fusion of Local and Global Features}, \n author={Min Yang and Dongliang He and Miao Fan and Baorong Shi and Xuetong Xue and Fu Li and Errui Ding and Jizhou Huang},\n year={2021},\n eprint={2108.02927},\n archivePrefix={arXiv},\n primaryClass={cs.CV}\n}\n\n\n@misc{https://doi.org/10.48550/arxiv.2004.01804,\n doi = {10.48550/ARXIV.2004.01804},\n\n url = {https://arxiv.org/abs/2004.01804},\n\n author = {Weyand, Tobias and Araujo, Andre and Cao, Bingyi and Sim, Jack},\n\n keywords = {Computer Vision and Pattern Recognition (cs.CV), FOS: Computer and information sciences, FOS: Computer and information sciences},\n\n title = {Google Landmarks Dataset v2 -- A Large-Scale Benchmark for Instance-Level Recognition and Retrieval},\n```\n\n",
"bugtrack_url": null,
"license": "MIT License",
"summary": "Re-implementation of DOLG paper in torch and tensorflow with converted checkpoints",
"version": "0.1.5.1",
"split_keywords": [
"dolg for torch and tensorflow",
"pretrained weights",
"tensorflow",
"tf",
"pytorch",
"torch"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "816d3a3ed16afda47e2b3869dfbf4f1e7dfb9fc9975577d08a26528d38923de6",
"md5": "b9e6fba410e1111c4247d4c12f311e4b",
"sha256": "588e31d5a1f3e34db456ef8aa98a858e60ed3609d0b9d29966b2c10b79095b27"
},
"downloads": -1,
"filename": "dolg-0.1.5.1-py3-none-any.whl",
"has_sig": false,
"md5_digest": "b9e6fba410e1111c4247d4c12f311e4b",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 18540,
"upload_time": "2023-01-02T23:46:45",
"upload_time_iso_8601": "2023-01-02T23:46:45.248032Z",
"url": "https://files.pythonhosted.org/packages/81/6d/3a3ed16afda47e2b3869dfbf4f1e7dfb9fc9975577d08a26528d38923de6/dolg-0.1.5.1-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "2a807ee6e1fa67130ac85949e1eab15e9bc4ebeb720c8f3d504f2fad97a3161e",
"md5": "cc4ee89a6d2ee64443df5ca12ab29fdb",
"sha256": "2cc4c6475479521c54853a55bedeb3da1437e6854def103a6fe58aca16471bd9"
},
"downloads": -1,
"filename": "dolg-0.1.5.1.tar.gz",
"has_sig": false,
"md5_digest": "cc4ee89a6d2ee64443df5ca12ab29fdb",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 15116,
"upload_time": "2023-01-02T23:46:56",
"upload_time_iso_8601": "2023-01-02T23:46:56.066860Z",
"url": "https://files.pythonhosted.org/packages/2a/80/7ee6e1fa67130ac85949e1eab15e9bc4ebeb720c8f3d504f2fad97a3161e/dolg-0.1.5.1.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-01-02 23:46:56",
"github": true,
"gitlab": false,
"bitbucket": false,
"github_user": "Shiro-LK",
"github_project": "python-DOLG",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "dolg"
}
JSON
