{
    "info": {
        "author": "Long Chen",
        "author_email": "looooong.chen@gmail.com",
        "bugtrack_url": null,
        "classifiers": [
            "License :: OSI Approved :: MIT License",
            "Operating System :: OS Independent",
            "Programming Language :: Python :: 3"
        ],
        "description": "\nstill under construction, coming soon\n# tfAugmentor\nAn image augmentation library for tensorflow. All operations are implemented as pure tensorflow graph operations. Thus, tfAugmentor can be easily combined with any tensorflow graph, such as tf.Data, for on-the-fly data augmentation. \nOf cousrse, you can also use it off-line to generate your augmented dataset. \n\n## Installation\ntfAugmentor is written in Python and can be easily installed via:\n```python\npip install tfAugmentor\n```\nTo run tfAugmentor properly, the following library should be installed as well:\n- tensorflow (developed under tf 1.12)\n- numpy (developed under numpy 1.15)\n\n## Quick Start\ntfAugmentor aims to implement image augmentations purly as a tensorflow graph, so that it can be used seamlessly with other tensorflow components, such as tf.Data. \nBut you can also use it independently as a off-line augmentation tool.   \n\nTo begin, instantiate an `Augmentor` object and pass a dictionary of tensors to it. These tensors should have the same 4-D shape of `[batch, height, width, channels]`. \n\nTo preserve the consistence of label/segmentation maps, the corresponding dict key should be pass to `label` as a list.\n\n```python\nimport tfAugmentor as tfa\ntensor_list = {\n\t'images': image_tensor,\n\t'segmentation_mask': mask_tensor\n}\naug1 = tfa.Augmentor(tensor_list, label=['segmentation_mask'])\n```\n\nAdd augmentations and get the output tensor:\n\n```python\naug1.flip_left_right(probability=0.5) # apply left right flip with probability 0.5\nout1 = aug1.out\n```\n\nSeveral augmentors with the same structure can be merged, which means you can parallel several pipelines\n\n```python\naug2 = tfa.Augmentor(tensor_list, label=['segmentation_mask']) # another augmentor with the same input as aug1\naug2.flip_up_down(probability=0.5) # apply up down flip this time\naug3 = tfa.Augmentor(tensor_list, label=['segmentation_mask'])\naug3.elastic_deform(probability=0.5, strength=1, scale=30) # elastic deformation\nout = aug1.merge([aug2, aug3])\n```\n\n### Example with tf.Data\n\nAn example of data importing with tf.data and tfAugmentor:\n\n```python\nds = tf.data.TFRecordDataset([...])\nds = ds.map(extract_fn)\nds = ds.shuffle(buffer_size=500)\nds = ds.batch(batch_size)\n\niterator = dataset.make_one_shot_iterator()\nnext_element = iterator.get_next()\n\n\ndef extract_fn(sample):\n\n\t# parse the tfrecord example of your dataset\n\t....\n\t# assume the dataset contains three tensors: image, weight_map, seg_mask\n\n\t# instantiate an Augmentor\n\tinput_list = {\n\t\t'img': image,\n\t\t'weight': weight_map,\n\t\t'mask': seg_mask\n\t}\n\ta = tfa.Augmentor(input_list, label=['segmentation_mask'])\n\n\ta.flip_left_right(probability=0.5) # apply left right flip\n\ta.random_rotate(probability=0.6) # apply random rotation\n\ta.elastic_deform(probability=0.2, strength=200, scale=20) # apply elastic deformation\n\n\t# dictionary of the augmented images, which has the same keys as input_list\n\taugmented = a.out\n\t# return tensors in a form you need\n\treturn augmented['img'], augmented['weight'], augmented['mask'] \n```\n\n## Main Features\n\n### Mirroring\n```python\na.flip_left_right(probability) # flip the image left right  \na.flip_up_down(probability) # flip the image up down\n```\n### Rotating\n```python\na.rotate90(probability) # rotate by 90 degree clockwise\na.rotate180(probability) # rotate by 180 degree clockwise\na.rotate270(probability) # rotate by 270 degree clockwise\na.rotate(probability, angle) # rotate by *angel* degree clockwise\na.random_rotate(probability) # randomly rotate the image\n```\n### crop and resize\n```python\na.random_crop_resize(probability, scale_range=(0.5, 0.8)) # randomly crop a sub-image and resize to the same size of the original image\na.crop(probability, size) # randomly crop a sub-image of a certain size\n```\n\n### elastic deformation\n```\na.elastic_deform(probability, strength, scale)\n```\n\n",
        "description_content_type": "text/markdown",
        "docs_url": null,
        "download_url": "",
        "downloads": {
            "last_day": -1,
            "last_month": -1,
            "last_week": -1
        },
        "home_page": "https://github.com/looooongChen/tfAugmentor",
        "keywords": "",
        "license": "",
        "maintainer": "",
        "maintainer_email": "",
        "name": "tfAugmentor",
        "package_url": "https://pypi.org/project/tfAugmentor/",
        "platform": "",
        "project_url": "https://pypi.org/project/tfAugmentor/",
        "project_urls": {
            "Homepage": "https://github.com/looooongChen/tfAugmentor"
        },
        "release_url": "https://pypi.org/project/tfAugmentor/1.0.2/",
        "requires_dist": [
            "tensorflow (>=1.12.0)",
            "numpy"
        ],
        "requires_python": "",
        "summary": "An image augmentation library for tensorflow. All operations are implemented as pure tensorflow graph operations.",
        "version": "1.0.2"
    },
    "last_serial": 5286928,
    "releases": {
        "1.0.1": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "0876a96a89699095df409b1b0c046edb",
                    "sha256": "e932f41bedbaffc6b409dcf4fcb64dec0873ca29bb1009528a35e56d665379ec"
                },
                "downloads": -1,
                "filename": "tfAugmentor-1.0.1-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "0876a96a89699095df409b1b0c046edb",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 7752,
                "upload_time": "2019-05-17T16:32:37",
                "url": "https://files.pythonhosted.org/packages/24/10/70e2ffb25407c190f82811bef81a70eb4ef1cf95eea7ba59891bef771394/tfAugmentor-1.0.1-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "1d994aac3709727773929412ca24ee05",
                    "sha256": "fa4efef64fe09d894d5e16e8b98157a618aadf856280439cfcac9e81d2c0145e"
                },
                "downloads": -1,
                "filename": "tfAugmentor-1.0.1.tar.gz",
                "has_sig": false,
                "md5_digest": "1d994aac3709727773929412ca24ee05",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 6870,
                "upload_time": "2019-05-17T16:32:39",
                "url": "https://files.pythonhosted.org/packages/0b/a8/a6e12a2537cd0975035011a44e6a6140c305943d0c29b5980f38aab3b251/tfAugmentor-1.0.1.tar.gz"
            }
        ],
        "1.0.2": [
            {
                "comment_text": "",
                "digests": {
                    "md5": "f19355e919b604a10ff75360a59608ac",
                    "sha256": "fbba479ea990a93a0f0984dea49915e358055a571ee37b4eb3e89115ad4182f0"
                },
                "downloads": -1,
                "filename": "tfAugmentor-1.0.2-py3-none-any.whl",
                "has_sig": false,
                "md5_digest": "f19355e919b604a10ff75360a59608ac",
                "packagetype": "bdist_wheel",
                "python_version": "py3",
                "requires_python": null,
                "size": 7982,
                "upload_time": "2019-05-19T00:12:50",
                "url": "https://files.pythonhosted.org/packages/4c/82/f4b80821d6eccead5c772d42b363257f6f59112326c5bc471370b0fe49db/tfAugmentor-1.0.2-py3-none-any.whl"
            },
            {
                "comment_text": "",
                "digests": {
                    "md5": "ba1ab1d2a9e7381e19ad0efa2464947c",
                    "sha256": "acfa4a022263e81cdff76fe227ed4ecb5aed95f44ee64a26a8e20d77b7ac573b"
                },
                "downloads": -1,
                "filename": "tfAugmentor-1.0.2.tar.gz",
                "has_sig": false,
                "md5_digest": "ba1ab1d2a9e7381e19ad0efa2464947c",
                "packagetype": "sdist",
                "python_version": "source",
                "requires_python": null,
                "size": 7161,
                "upload_time": "2019-05-19T00:12:52",
                "url": "https://files.pythonhosted.org/packages/60/f9/dd2484a630a441135b7bdd0886aaa25dd355fea0df731d69f879932c55d8/tfAugmentor-1.0.2.tar.gz"
            }
        ]
    },
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "f19355e919b604a10ff75360a59608ac",
                "sha256": "fbba479ea990a93a0f0984dea49915e358055a571ee37b4eb3e89115ad4182f0"
            },
            "downloads": -1,
            "filename": "tfAugmentor-1.0.2-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "f19355e919b604a10ff75360a59608ac",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": null,
            "size": 7982,
            "upload_time": "2019-05-19T00:12:50",
            "url": "https://files.pythonhosted.org/packages/4c/82/f4b80821d6eccead5c772d42b363257f6f59112326c5bc471370b0fe49db/tfAugmentor-1.0.2-py3-none-any.whl"
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "ba1ab1d2a9e7381e19ad0efa2464947c",
                "sha256": "acfa4a022263e81cdff76fe227ed4ecb5aed95f44ee64a26a8e20d77b7ac573b"
            },
            "downloads": -1,
            "filename": "tfAugmentor-1.0.2.tar.gz",
            "has_sig": false,
            "md5_digest": "ba1ab1d2a9e7381e19ad0efa2464947c",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": null,
            "size": 7161,
            "upload_time": "2019-05-19T00:12:52",
            "url": "https://files.pythonhosted.org/packages/60/f9/dd2484a630a441135b7bdd0886aaa25dd355fea0df731d69f879932c55d8/tfAugmentor-1.0.2.tar.gz"
        }
    ]
}