{
    "info": {
        "author": "Manuel Schoelling, Christoph Moehl",
        "author_email": "manuel.schoelling@dzne.de, christoph.moehl@dzne.de",
        "bugtrack_url": null,
        "classifiers": [],
        "description": "[![Build Status](https://travis-ci.com/yapic/yapic_io.svg?branch=master)](https://travis-ci.com/yapic/yapic_io)\n[![Documentation Status](https://readthedocs.org/projects/yapic-io/badge/?version=latest)](https://yapic-io.readthedocs.io/en/latest/?badge=latest)\n\n# yapic_io\n\n\n\n\nyapic_io provides flexible data binding to image collections of arbitrary size.\n\n\nIts aim is to provide a convenient image data interface for training of\nfully convolutional neural networks, as well as automatic handling of\nprediction data output for a trained classifier.\n\nyapic_io is designed as a convenient image data input/output interface for\nlibraries such as Theano or TensorFlow.\n\n\nFollowing problems occuring with training/classification are handeled by yapic_io:\n\n- Images of different sizes in z,x, and y can be applied to the\n  same convolutional network. This is implemented by sliding windows. The size these windows correspond to the size of the convolutional network's input layer.\n\n- Due to lazy data loading, images can be extremely large.\n\n- Image dimensions can be up to 4D (multchannel z-stack), as e.g. required\n  for bioimages.\n\n- Data augmentation for classifier training in built in.\n\n- Made for sparsly labelled datasets: Training data is only (randomly) picked\n  from regions where labels are present.\n\n- Usually, input layers of CNNs are larger than output layers. Thus, pixels\n  located at image edges are normally not classified. With yapic_io also\n  edge pixels are classified. This is achieved by mirroring pixel data in edge\n  regions. As a result, output classification images have identical dimensions as source images and can be overlayed easily.\n\n\n\n## Example\n\nClassifier training:\n\n```\n>>> from yapic_io import TiffConnector, Dataset, TrainingBatch\n>>>\n>>> #define data locations\n>>> pixel_image_dir = 'yapic_io/test_data/tiffconnector_1/im/*.tif'\n>>> label_image_dir = 'yapic_io/test_data/tiffconnector_1/labels/*.tif'\n>>> savepath = 'yapic_io/test_data/tmp/'\n>>>\n>>>\n>>> tpl_size = (1,5,4) # size of network output layer in zxy\n>>> padding = (0,2,2) # padding of network input layer in zxy, in respect to output layer\n>>>\n>>> c = TiffConnector(pixel_image_dir, label_image_dir, savepath=savepath)\n>>> train_data = TrainingBatch(Dataset(c), tpl_size, padding_zxy=padding)\n>>>\n>>> counter=0\n>>> for mini in train_data:\n...     weights = mini.weights\n...     #shape of weights is (6,3,1,5,4) : batchsize 6 , 3 label-classes, 1 z, 5 x, 4 y\n...\n...     pixels = mini.pixels()\n...     # shape of pixels is (6,3,1,9,8) : 3 channels, 1 z, 9 x, 4 y (more xy due to padding)\n...\n...     #here: apply training on mini.pixels and mini.weights (use theano, tensorflow...)\n...     my_train_function(pixels, weights)\n...\n...     counter += 1\n...     if counter > 10: #m is infinite\n...         break\n```\nPrediction:\n```\n>>> from yapic_io import TiffConnector, Dataset, PredictionBatch\n>>>\n>>> #mock classification function\n>>> def classify(pixels, value):\n...     return np.ones(pixels.shape) * value\n>>>\n>>> #define data loacations\n>>> pixel_image_dir = 'yapic_io/test_data/tiffconnector_1/im/*.tif'\n>>> label_image_dir = 'yapic_io/test_data/tiffconnector_1/labels/*.tif'\n>>> savepath = 'yapic_io/test_data/tmp/'\n>>>\n>>> tpl_size = (1,5,4) # size of network output layer in zxy\n>>> padding = (0,2,2) # padding of network input layer in zxy, in respect to output layer\n>>>\n>>> c = TiffConnector(pixel_image_dir, label_image_dir, savepath=savepath)\n>>> prediction_data = PredictionBatch(Dataset(c))\n>>> len(prediction_data) #give the total number of templates that cover the whole bound tifffiles\n510\n>>>\n>>> #classify the whole bound dataset\n>>> counter = 0 #needed for mock data\n>>> for item in prediction_data:\n...     pixels_for_classifier = item.pixels() #input for classifier\n...     mock_classifier_result = classify(pixels, counter) #classifier output\n...\n...     #pass classifier results for each class to data source\n...     item.put_probmap_data(mock_classifier_result)\n...\n...     counter += 1 #counter for generation of mockdata\n>>>\n```\n\n## Buils API docs\n\n```\ncd docs\nsphinx-apidoc -o source ../yapic_io\nmake html\n```\n\n\nDeveloped by the CRFS (Core Research Facilities) of the DZNE (German Center\nfor Neurodegenerative Diseases).",
        "description_content_type": "",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "",
        "keywords": "",
        "license": "",
        "maintainer": "",
        "maintainer_email": "",
        "name": "yapic-io",
        "package_url": "https://pypi.org/project/yapic-io/",
        "platform": "",
        "project_url": "https://pypi.org/project/yapic-io/",
        "project_urls": null,
        "release_url": "https://pypi.org/project/yapic-io/0.1.0/",
        "requires_dist": null,
        "requires_python": "",
        "summary": "io data handling module for various image sources as interface for pixel classification tools",
        "version": "0.1.0"
    },
    "last_serial": 5272694,
    "releases": {
        "0.1.0": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "ef45a06b8330f920f6d1791dc9333d3f",
                    "sha256": "e2759f2066820ef60266ae98e1d384e5cb0230f522bd071971b4e2e6b4d68197"
                },
                "downloads": -1,
                "filename": "yapic_io-0.1.0.tar.gz",
                "has_sig": false,
                "md5_digest": "ef45a06b8330f920f6d1791dc9333d3f",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 24718,
                "upload_time": "2019-05-15T14:12:01",
                "url": "https://files.pythonhosted.org/packages/2f/63/dd7953436f97a6237c9a23783429b14d28648f487108143972e6aa32598d/yapic_io-0.1.0.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "ef45a06b8330f920f6d1791dc9333d3f",
                "sha256": "e2759f2066820ef60266ae98e1d384e5cb0230f522bd071971b4e2e6b4d68197"
            },
            "downloads": -1,
            "filename": "yapic_io-0.1.0.tar.gz",
            "has_sig": false,
            "md5_digest": "ef45a06b8330f920f6d1791dc9333d3f",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 24718,
            "upload_time": "2019-05-15T14:12:01",
            "url": "https://files.pythonhosted.org/packages/2f/63/dd7953436f97a6237c9a23783429b14d28648f487108143972e6aa32598d/yapic_io-0.1.0.tar.gz"
        }
    ]
}