# SemTorch
This repository contains different deep learning architectures definitions that can be applied to image segmentation.
All the architectures are implemented in [PyTorch](https://pytorch.org/) and can been trained easily with [FastAI 2](https://github.com/fastai/fastaihttps://github.com/fastai/fastai).
In [Deep-Tumour-Spheroid repository](https://github.com/WaterKnight1998/Deep-Tumour-Spheroid) can be found and example of how to apply it with a custom dataset, in that case brain tumours images are used.
These architectures are classified as:
* **Semantic Segmentation:** each pixel of an image is linked to a class label.
* **Instance Segmentation:** is similar to semantic segmentation, but goes a bit deeper, it identifies , for each pixel, the object instance it belongs to.
* **Salient Object Detection (Binary clases only):** detection of the most noticeable/important object in an image.
## 🚀 Getting Started
To start using this package, install it using `pip`:
For example, for installing it in Ubuntu use:
```bash
pip3 install SemTorch
```
## 👩💻 Usage
This package creates an abstract API to access a segmentation model of different architectures. This method returns a FastAI 2 learner that can be combined with all the fastai's functionalities.
```
# SemTorch
from semtorch import get_segmentation_learner
learn = get_segmentation_learner(dls=dls, number_classes=2, segmentation_type="Semantic Segmentation",
architecture_name="deeplabv3+", backbone_name="resnet50",
metrics=[tumour, Dice(), JaccardCoeff()],wd=1e-2,
splitter=segmentron_splitter).to_fp16()
```
You can find a deeper example in [Deep-Tumour-Spheroid repository](https://github.com/WaterKnight1998/Deep-Tumour-Spheroid/tree/feature/notebooks), in this repo the package is used for the segmentation of brain tumours.
```
def get_segmentation_learner(dls, number_classes, segmentation_type, architecture_name, backbone_name,
loss_func=None, opt_func=Adam, lr=defaults.lr, splitter=trainable_params,
cbs=None, pretrained=True, normalize=True, image_size=None, metrics=None,
path=None, model_dir='models', wd=None, wd_bn_bias=False, train_bn=True,
moms=(0.95,0.85,0.95)):
```
This function return a learner for the provided architecture and backbone
### **Parameters:**
* **dls (DataLoader):** the dataloader to use with the learner
* **number_classes (int):** the number of clases in the project. It should be >=2
* **segmentation_type (str):** just `Semantic Segmentation` accepted for now
* **architecture_name (str):** name of the architecture. The following ones are supported: `unet`, `deeplabv3+`, `hrnet`, `maskrcnn` and `u2^net`
* **backbone_name (str):** name of the backbone
* **loss_func ():** loss function.
* **opt_func ():** opt function.
* **lr ():** learning rates
* **splitter ():** splitter function for freazing the learner
* **cbs (List[cb]):** list of callbacks
* **pretrained (bool):** it defines if a trained backbone is needed
* **normalize (bool):** if normalization is applied
* **image_size (int):** REQUIRED for MaskRCNN. It indicates the desired size of the image.
* **metrics (List[metric]):** list of metrics
* **path ():** path parameter
* **model_dir (str):** the path in which save models
* **wd (float):** wieght decay
* **wd_bn_bias (bool):**
* **train_bn (bool):**
* **moms (Tuple(float)):** tuple of different momentuns
### **Returns:**
* **learner:** value containing the learner object
### **Supported configs**
| Architecture | supported config | backbones |
|--------------|:--------------------------------------------------------------------:|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------:|
| unet | `Semantic Segmentation`,`binary` `Semantic Segmentation`,`multiple` | `resnet18`, `resnet34`, `resnet50`, `resnet101`, `resnet152`, `xresnet18`, `xresnet34`, `xresnet50`, `xresnet101`, `xresnet152`, `squeezenet1_0`, `squeezenet1_1`, `densenet121`, `densenet169`, `densenet201`, `densenet161`, `vgg11_bn`, `vgg13_bn`, `vgg16_bn`, `vgg19_bn`, `alexnet` |
| deeplabv3+ | `Semantic Segmentation`,`binary` `Semantic Segmentation`,`multiple` | `resnet18`, `resnet34`, `resnet50`, `resnet101`, `resnet152`, `resnet50c`, `resnet101c`, `resnet152c`, `xception65`, `mobilenet_v2` |
| hrnet | `Semantic Segmentation`,`binary` `Semantic Segmentation`,`multiple` | `hrnet_w18_small_model_v1`, `hrnet_w18_small_model_v2`, `hrnet_w18`, `hrnet_w30`, `hrnet_w32`, `hrnet_w48` |
| maskrcnn | `Semantic Segmentation`,`binary` | `resnet50` |
| u2^net | `Semantic Segmentation`,`binary` | `small`, `normal` |
## 📩 Contact
📧 dvdlacallecastillo@gmail.com
💼 Linkedin [David Lacalle Castillo](https://es.linkedin.com/in/david-lacalle-castillo-5b6280173)
Raw data
{
"_id": null,
"home_page": "https://github.com/WaterKnight1998/SemTorch",
"name": "SemTorch",
"maintainer": "David Lacalle Castillo",
"docs_url": null,
"requires_python": "",
"maintainer_email": "dvdlacallecastillo@gmail.com",
"keywords": "instance semantic segmentation pytorch fastai fastai2 fastaiv2 saliend object detection",
"author": "David Lacalle Castillo",
"author_email": "dvdlacallecastillo@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/78/59/e41afbd4cf5f8bcbf6d0b8117f60586e87f9053610fcb262d2af950ab7b4/SemTorch-0.1.1.tar.gz",
"platform": "",
"description": "# SemTorch\n\nThis repository contains different deep learning architectures definitions that can be applied to image segmentation. \n\nAll the architectures are implemented in [PyTorch](https://pytorch.org/) and can been trained easily with [FastAI 2](https://github.com/fastai/fastaihttps://github.com/fastai/fastai). \n\nIn [Deep-Tumour-Spheroid repository](https://github.com/WaterKnight1998/Deep-Tumour-Spheroid) can be found and example of how to apply it with a custom dataset, in that case brain tumours images are used.\n\nThese architectures are classified as:\n\n* **Semantic Segmentation:** each pixel of an image is linked to a class label.\n\n* **Instance Segmentation:** is similar to semantic segmentation, but goes a bit deeper, it identifies , for each pixel, the object instance it belongs to.\n\n* **Salient Object Detection (Binary clases only):** detection of the most noticeable/important object in an image.\n\n\n## \ud83d\ude80 Getting Started\n\nTo start using this package, install it using `pip`:\n\nFor example, for installing it in Ubuntu use:\n```bash\npip3 install SemTorch\n```\n\n## \ud83d\udc69\u200d\ud83d\udcbb Usage\nThis package creates an abstract API to access a segmentation model of different architectures. This method returns a FastAI 2 learner that can be combined with all the fastai's functionalities.\n\n```\n# SemTorch\nfrom semtorch import get_segmentation_learner\n\nlearn = get_segmentation_learner(dls=dls, number_classes=2, segmentation_type=\"Semantic Segmentation\",\n architecture_name=\"deeplabv3+\", backbone_name=\"resnet50\", \n metrics=[tumour, Dice(), JaccardCoeff()],wd=1e-2,\n splitter=segmentron_splitter).to_fp16()\n```\n\nYou can find a deeper example in [Deep-Tumour-Spheroid repository](https://github.com/WaterKnight1998/Deep-Tumour-Spheroid/tree/feature/notebooks), in this repo the package is used for the segmentation of brain tumours.\n\n```\ndef get_segmentation_learner(dls, number_classes, segmentation_type, architecture_name, backbone_name,\n loss_func=None, opt_func=Adam, lr=defaults.lr, splitter=trainable_params, \n cbs=None, pretrained=True, normalize=True, image_size=None, metrics=None, \n path=None, model_dir='models', wd=None, wd_bn_bias=False, train_bn=True,\n moms=(0.95,0.85,0.95)):\n```\n\nThis function return a learner for the provided architecture and backbone\n\n### **Parameters:**\n\n* **dls (DataLoader):** the dataloader to use with the learner\n* **number_classes (int):** the number of clases in the project. It should be >=2\n* **segmentation_type (str):** just `Semantic Segmentation` accepted for now \n* **architecture_name (str):** name of the architecture. The following ones are supported: `unet`, `deeplabv3+`, `hrnet`, `maskrcnn` and `u2^net`\n* **backbone_name (str):** name of the backbone\n* **loss_func ():** loss function.\n* **opt_func ():** opt function.\n* **lr ():** learning rates\n* **splitter ():** splitter function for freazing the learner\n* **cbs (List[cb]):** list of callbacks\n* **pretrained (bool):** it defines if a trained backbone is needed\n* **normalize (bool):** if normalization is applied\n* **image_size (int):** REQUIRED for MaskRCNN. It indicates the desired size of the image.\n* **metrics (List[metric]):** list of metrics\n* **path ():** path parameter\n* **model_dir (str):** the path in which save models\n* **wd (float):** wieght decay\n* **wd_bn_bias (bool):**\n* **train_bn (bool):**\n* **moms (Tuple(float)):** tuple of different momentuns\n\n### **Returns:**\n\n* **learner:** value containing the learner object\n\n### **Supported configs**\n\n| Architecture | supported config | backbones |\n|--------------|:--------------------------------------------------------------------:|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------:|\n| unet | `Semantic Segmentation`,`binary` `Semantic Segmentation`,`multiple` | `resnet18`, `resnet34`, `resnet50`, `resnet101`, `resnet152`, `xresnet18`, `xresnet34`, `xresnet50`, `xresnet101`, `xresnet152`, `squeezenet1_0`, `squeezenet1_1`, `densenet121`, `densenet169`, `densenet201`, `densenet161`, `vgg11_bn`, `vgg13_bn`, `vgg16_bn`, `vgg19_bn`, `alexnet` |\n| deeplabv3+ | `Semantic Segmentation`,`binary` `Semantic Segmentation`,`multiple` | `resnet18`, `resnet34`, `resnet50`, `resnet101`, `resnet152`, `resnet50c`, `resnet101c`, `resnet152c`, `xception65`, `mobilenet_v2` |\n| hrnet | `Semantic Segmentation`,`binary` `Semantic Segmentation`,`multiple` | `hrnet_w18_small_model_v1`, `hrnet_w18_small_model_v2`, `hrnet_w18`, `hrnet_w30`, `hrnet_w32`, `hrnet_w48` |\n| maskrcnn | `Semantic Segmentation`,`binary` | `resnet50` |\n| u2^net | `Semantic Segmentation`,`binary` | `small`, `normal` |\n\n## \ud83d\udce9 Contact\n\ud83d\udce7 dvdlacallecastillo@gmail.com\n\n\ud83d\udcbc Linkedin [David Lacalle Castillo](https://es.linkedin.com/in/david-lacalle-castillo-5b6280173)",
"bugtrack_url": null,
"license": "Apache License Version 2.0",
"summary": "Deep Leaarning segmentation architectures for PyTorch and FastAI",
"version": "0.1.1",
"project_urls": {
"Homepage": "https://github.com/WaterKnight1998/SemTorch"
},
"split_keywords": [
"instance",
"semantic",
"segmentation",
"pytorch",
"fastai",
"fastai2",
"fastaiv2",
"saliend",
"object",
"detection"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "7859e41afbd4cf5f8bcbf6d0b8117f60586e87f9053610fcb262d2af950ab7b4",
"md5": "f4ae1f6ec463150b9565f3dbed53d42a",
"sha256": "f0814cd387fc03581fbff4a63b63a50dc889e19780b52273d5a734328c5f1647"
},
"downloads": -1,
"filename": "SemTorch-0.1.1.tar.gz",
"has_sig": false,
"md5_digest": "f4ae1f6ec463150b9565f3dbed53d42a",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 41795,
"upload_time": "2020-09-08T21:49:32",
"upload_time_iso_8601": "2020-09-08T21:49:32.764124Z",
"url": "https://files.pythonhosted.org/packages/78/59/e41afbd4cf5f8bcbf6d0b8117f60586e87f9053610fcb262d2af950ab7b4/SemTorch-0.1.1.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2020-09-08 21:49:32",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "WaterKnight1998",
"github_project": "SemTorch",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"requirements": [],
"lcname": "semtorch"
}
JSON
