# SIQ - super-resolution image quantification
## deep perceptual resampling and super-resolution for (medical) imaging
install by calling (within the source directory):
```
python setup.py install
```
or install via `pip install siq`
# what this will do
facilitates:
* creating training and testing data for deep networks
* generating and testing perceptual losses in 2D and 3D
* general training and inference functions for deep networks
* intuitive weighting of multiple losses
* anisotropic super-resolution
* evaluation strategies for the above
# first time setup
```python
import antspyt1w
antspyt1w.get_data( force_download=True )
# import siq # FIXME - for later
# siq.get_data( force_download=True )
```
NOTE: `get_data` has a `force_download` option to make sure the latest
package data is installed.
# example processing
```python
import os
import siq
import glob
import ants
fns=glob.glob( os.path.expanduser( "~/.antspyt1w/2*T1w*gz" ) )
import tensorflow as tf
ofn = os.path.expanduser("~/code/DPR/models/dsr3d_2up_64_256_6_3_v0.0zzz.h5")
if os.path.exists( ofn ):
print("existing model") # should always initialize with pre-trained model
mdl = tf.keras.models.load_model( ofn, compile=False )
else:
print("default model - initialized with random weights")
mdl = siq.default_dbpn( [2,2,2] ) # should match ratio of high to low size patches
myoutprefix = '/tmp/XXX'
training_path = siq.train(
mdl,
fns[0:3],
fns[0:3],
output_prefix=myoutprefix,
target_patch_size=[32,32,32],
target_patch_size_low=[16,16,16],
n_test=2,
learning_rate=5e-05,
feature_layer=6,
feature=2,
tv=0.1,
max_iterations=2,
verbose=True)
training_path.to_csv( myoutprefix + "_training.csv" )
image = ants.image_read( fns[0] )
image = ants.resample_image( image, [48,48,48] ) # downsample for speed in testing
test = siq.inference( image, mdl )
```
see also: the training scripts in `tests`.
## todo
1. numpy read/write
2. test/fix 2D
## your compute environment
```bash
export TF_ENABLE_ONEDNN_OPTS=1 # for CPU
total_cpu_cores=$(nproc)
number_sockets=$(($(grep "^physical id" /proc/cpuinfo | awk '{print $4}' | sort -un | tail -1)+1))
number_cpu_cores=$(( (total_cpu_cores/2) / number_sockets))
echo "number of CPU cores per socket: $number_cpu_cores";
echo "number of socket: $number_sockets";
echo "Physical cores:"
egrep '^core id' /proc/cpuinfo | sort -u | wc -l
echo "Logical cores:"
egrep '^processor' /proc/cpuinfo | sort -u | wc -l
echo "Physical cpus (separate chips):"
egrep '^physical id' /proc/cpuinfo | sort -u | wc -l
```
## to publish a release
```
rm -r -f build/ antspymm.egg-info/ dist/
python3 setup.py sdist bdist_wheel
python3 -m twine upload -u username -p password dist/*
```
## notes on cpu environment
```
# dd=/home/ubuntu/miniconda3/condabin/conda
# conda update -n base -c defaults conda
# conda init bash
# conda create -n ai3 python=3.9
# conda activate ai3
# pip3 install --upgrade pip
py=python3 # "sudo /opt/parallelcluster/pyenv/versions/3.7.10/envs/awsbatch_virtualenv/bin/python3.7"
$py -m pip install --upgrade pip
# python3.7 -m pip uninstall tensorflow antspynet dipy patsy tensorboard tensorflow-probability -y
$py -m pip install nibabel PyNomaly scipy
$py -m pip install antspyx
$py -m pip install dipy
$py -m pip install antspyt1w
$py -m pip install antspymm
$py -m pip install antspynet
$py -m pip install siq
$py -m pip uninstall tensorflow -y
$py -m pip install intel-tensorflow # -avx512==2.9.1
$py -m pip install tensorflow_probability
$py -m pip install keras
```
Raw data
{
"_id": null,
"home_page": "https://github.com/stnava/siq",
"name": "siq",
"maintainer": "",
"docs_url": null,
"requires_python": "",
"maintainer_email": "",
"keywords": "",
"author": "Avants, Gosselin, Tustison, Reardon",
"author_email": "stnava@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/ac/e8/7ff42a4497f42702b1d733ae82b693e8fa47e0bdb7a9da9ade4a683e8604/siq-0.3.4.tar.gz",
"platform": null,
"description": "# SIQ - super-resolution image quantification\n\n## deep perceptual resampling and super-resolution for (medical) imaging\n\ninstall by calling (within the source directory):\n\n```\npython setup.py install\n```\n\nor install via `pip install siq`\n\n# what this will do\n\nfacilitates:\n\n* creating training and testing data for deep networks\n\n* generating and testing perceptual losses in 2D and 3D\n\n* general training and inference functions for deep networks\n\n* intuitive weighting of multiple losses\n\n* anisotropic super-resolution\n\n* evaluation strategies for the above\n\n# first time setup\n\n```python\nimport antspyt1w\nantspyt1w.get_data( force_download=True )\n# import siq # FIXME - for later\n# siq.get_data( force_download=True )\n```\n\nNOTE: `get_data` has a `force_download` option to make sure the latest\npackage data is installed.\n\n# example processing\n\n```python\nimport os\nimport siq\nimport glob\nimport ants\nfns=glob.glob( os.path.expanduser( \"~/.antspyt1w/2*T1w*gz\" ) )\nimport tensorflow as tf\nofn = os.path.expanduser(\"~/code/DPR/models/dsr3d_2up_64_256_6_3_v0.0zzz.h5\")\nif os.path.exists( ofn ):\n print(\"existing model\") # should always initialize with pre-trained model\n mdl = tf.keras.models.load_model( ofn, compile=False )\nelse:\n print(\"default model - initialized with random weights\")\n mdl = siq.default_dbpn( [2,2,2] ) # should match ratio of high to low size patches\nmyoutprefix = '/tmp/XXX'\ntraining_path = siq.train(\n mdl, \n fns[0:3], \n fns[0:3], \n output_prefix=myoutprefix,\n target_patch_size=[32,32,32],\n target_patch_size_low=[16,16,16],\n n_test=2, \n learning_rate=5e-05, \n feature_layer=6, \n feature=2, \n tv=0.1,\n max_iterations=2, \n verbose=True)\ntraining_path.to_csv( myoutprefix + \"_training.csv\" )\nimage = ants.image_read( fns[0] )\nimage = ants.resample_image( image, [48,48,48] ) # downsample for speed in testing\ntest = siq.inference( image, mdl )\n```\n\nsee also: the training scripts in `tests`.\n\n## todo\n\n1. numpy read/write\n\n2. test/fix 2D\n\n## your compute environment\n\n```bash\nexport TF_ENABLE_ONEDNN_OPTS=1 # for CPU\n\ntotal_cpu_cores=$(nproc)\nnumber_sockets=$(($(grep \"^physical id\" /proc/cpuinfo | awk '{print $4}' | sort -un | tail -1)+1))\nnumber_cpu_cores=$(( (total_cpu_cores/2) / number_sockets))\n\necho \"number of CPU cores per socket: $number_cpu_cores\";\necho \"number of socket: $number_sockets\";\n\necho \"Physical cores:\"\negrep '^core id' /proc/cpuinfo | sort -u | wc -l\n\necho \"Logical cores:\"\n\negrep '^processor' /proc/cpuinfo | sort -u | wc -l\n\necho \"Physical cpus (separate chips):\"\n\negrep '^physical id' /proc/cpuinfo | sort -u | wc -l\n\n```\n\n## to publish a release\n\n```\nrm -r -f build/ antspymm.egg-info/ dist/\npython3 setup.py sdist bdist_wheel\npython3 -m twine upload -u username -p password dist/*\n```\n\n\n## notes on cpu environment\n\n```\n# dd=/home/ubuntu/miniconda3/condabin/conda\n# conda update -n base -c defaults conda\n# conda init bash\n# conda create -n ai3 python=3.9\n# conda activate ai3 \n# pip3 install --upgrade pip\npy=python3 # \"sudo /opt/parallelcluster/pyenv/versions/3.7.10/envs/awsbatch_virtualenv/bin/python3.7\"\n\n$py -m pip install --upgrade pip\n\n# python3.7 -m pip uninstall tensorflow antspynet dipy patsy tensorboard tensorflow-probability -y\n$py -m pip install nibabel PyNomaly scipy \n$py -m pip install antspyx \n$py -m pip install dipy \n$py -m pip install antspyt1w \n$py -m pip install antspymm \n$py -m pip install antspynet\n$py -m pip install siq\n$py -m pip uninstall tensorflow -y\n$py -m pip install intel-tensorflow # -avx512==2.9.1\n$py -m pip install tensorflow_probability\n$py -m pip install keras\n```\n",
"bugtrack_url": null,
"license": "Apache 2.0",
"summary": "deep perceptual resampling and super resolution with antspyx",
"version": "0.3.4",
"project_urls": {
"Homepage": "https://github.com/stnava/siq"
},
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "2627b56e38f46478eeda80a30a4e547b37506dac71dc680f08b233207e117721",
"md5": "00041b8712c767c97b5c1ad48b6add05",
"sha256": "b03513cf67d393a1014be54595bbf055c804965929adf630e6cb0508250420bb"
},
"downloads": -1,
"filename": "siq-0.3.4-py3-none-any.whl",
"has_sig": false,
"md5_digest": "00041b8712c767c97b5c1ad48b6add05",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 19184,
"upload_time": "2023-08-21T12:29:28",
"upload_time_iso_8601": "2023-08-21T12:29:28.454985Z",
"url": "https://files.pythonhosted.org/packages/26/27/b56e38f46478eeda80a30a4e547b37506dac71dc680f08b233207e117721/siq-0.3.4-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "ace87ff42a4497f42702b1d733ae82b693e8fa47e0bdb7a9da9ade4a683e8604",
"md5": "4bf807e4826cad99b1f2b6beef9483f2",
"sha256": "2ab8ddfa16d4e4a4cb1aaee83365c5af800353bbaa9027ba2fda2b2ac297ffc8"
},
"downloads": -1,
"filename": "siq-0.3.4.tar.gz",
"has_sig": false,
"md5_digest": "4bf807e4826cad99b1f2b6beef9483f2",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 20743,
"upload_time": "2023-08-21T12:29:29",
"upload_time_iso_8601": "2023-08-21T12:29:29.576483Z",
"url": "https://files.pythonhosted.org/packages/ac/e8/7ff42a4497f42702b1d733ae82b693e8fa47e0bdb7a9da9ade4a683e8604/siq-0.3.4.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-08-21 12:29:29",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "stnava",
"github_project": "siq",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"requirements": [
{
"name": "h5py",
"specs": [
[
">=",
"2.10.0"
]
]
},
{
"name": "numpy",
"specs": [
[
">=",
"1.19.4"
]
]
},
{
"name": "pandas",
"specs": [
[
">=",
"1.0.1"
]
]
},
{
"name": "antspyx",
"specs": [
[
">=",
"0.2.7"
]
]
},
{
"name": "antspynet",
"specs": [
[
">=",
"0.0"
]
]
},
{
"name": "antspyt1w",
"specs": [
[
">=",
"0.2.3"
]
]
},
{
"name": "tensorflow",
"specs": [
[
">=",
"2.2"
]
]
},
{
"name": "pathlib",
"specs": []
}
],
"lcname": "siq"
}
JSON
