Retina-VesselNet-master
所属分类:其他
开发工具:Python
文件大小:25KB
下载次数:2
上传日期:2019-12-30 15:05:05
上 传 者:
chengzk
说明: Python版本的眼底血管提取例程,适用于图像信号处理课程项目
(Python version of fundus blood vessel extraction routine, applicable to the course project of image signal processing)
文件列表:
configs (0, 2019-04-11)
configs\segmention_config.json (426, 2019-04-11)
configs\utils (0, 2019-04-11)
configs\utils\__init__.py (107, 2019-04-11)
configs\utils\config_utils.py (2075, 2019-04-11)
configs\utils\img_utils.py (9892, 2019-04-11)
configs\utils\utils.py (4177, 2019-04-11)
experiments (0, 2019-04-11)
experiments\data_loaders (0, 2019-04-11)
experiments\data_loaders\__init__.py (107, 2019-04-11)
experiments\data_loaders\standard_loader.py (3969, 2019-04-11)
main_test.py (988, 2019-04-11)
main_train.py (1347, 2019-04-11)
perception (0, 2019-04-11)
perception\bases (0, 2019-04-11)
perception\bases\__init__.py (86, 2019-04-11)
perception\bases\data_loader_base.py (608, 2019-04-11)
perception\bases\infer_base.py (440, 2019-04-11)
perception\bases\model_base.py (1214, 2019-04-11)
perception\bases\trainer_base.py (409, 2019-04-11)
perception\infers (0, 2019-04-11)
perception\infers\__init__.py (86, 2019-04-11)
perception\infers\segmention_infer.py (2014, 2019-04-11)
perception\metric (0, 2019-04-11)
perception\metric\__init__.py (86, 2019-04-11)
perception\metric\segmention_metric.py (6216, 2019-04-11)
perception\models (0, 2019-04-11)
perception\models\__init__.py (86, 2019-04-11)
perception\models\dense_unet.py (3773, 2019-04-11)
perception\models\segmention_model.py (5178, 2019-04-11)
perception\trainers (0, 2019-04-11)
perception\trainers\__init__.py (107, 2019-04-11)
perception\trainers\segmention_trainer.py (5938, 2019-04-11)
root_dir.py (190, 2019-04-11)
#### you can find here
- [VesselNet](#vesselnet)
- [About Model](#about-model)
- [Mertic](#mertic)
- [Project Structure](#project-structure)
- [First to run](#first-to-run)
- [Pretrained Model](#pretrained-model)
- [Test your own image](#test-your-own-image)
- [Reference](#reference)
- [Future Work](#future-work)
# VesselNet
A DenseBlock-Unet for Retinal Blood Vessel Segmentation
**Notice:This Project structure updated on 9th June!**
You can find old version in *branch old*
![TestResult](https://i.imgur.com/pPMANyZ.jpg)
## About Model
This model is inspired by DenseNet and [@orobix/retina-unet][5], I modify the Conv2d block to DenseBlock and finally I get better result. The DenseBlock struct is shown below. This struct maximisely use the extracted feature. If u want further information, please read the [DenseNet Paper][3] and [code][4]
![DenseBlock](https://i.imgur.com/E2fDtOm.png)
## Result Evaluation
Tried With 40 images of DRIVE dataset and DenseBlock-Unet model.
Results on DRIVE database:
|Methods|AUC ROC on DRIVE|
|-:|-:|
|Liskowski|0.9790|
|Retina-Unet|0.9790|
|VesselNet|0.***41|
## Project Structure
The structure is based on my own [DL_Segmention_Template][1]. Difference between this project and the template is that we have metric module in dir: `perception/metric/`. To get more Information about the structure please see readme in [DL_Segmention_Template][1].
You can find model parameter in **configs/segmention_config.json**.
### First to run
**please run main_trainer.py first time**, then you will get data_route in experiment dir. Put your data in there, now you can run `main_trainer.py` again to train a model.
### Pretrained Model
The model is trained with *DRIVE dataset* on my own desktop (intel i7-7700hq,24g,gtx1050 2g) within 30 minutes.
Dataset and pretrained model can be found [here][2]. For Chinese, you can download [here][6].
### Test your own image
If u want to test your own image, put ur image to **(VesselNet)/test/origin**, and change the img_type of predict settings in **configs/segmention_config.json**, run `main_test.py` to get your result. The result is in **(VesselNet)/test/result**
## Reference
This project is based on the following 2 papers:
[U-Net: Convolutional Networks for Biomedical Image Segmentation](8)
[Densely Connected Convolutional Networks](7)
## Future Work
First of all, I choose 48x48pix patches to train the model. The patch size means that model can't be too deep. So in future, I want to test 128X128pix patches and 96x96 patches.
Second, Attention-based Unet and DeepLab-v3+ are also worth to try.
[1]: https://github.com/DeepTrial/DL_Segmention_Template
[2]: https://drive.google.com/file/d/1RALItn7a-XIe-ebsghk6HL-T0btJI9w7/view?usp=sharing
[3]: https://arxiv.org/pdf/1608.06993.pdf
[4]: https://github.com/liuzhuang13/DenseNet
[5]: https://github.com/orobix/retina-unet
[6]: https://pan.baidu.com/s/1EnKeNTGimzVRa9QedWjxlg
[7]: https://arxiv.org/pdf/1608.06993.pdf
[8]: https://arxiv.org/pdf/1505.04597.pdf
近期下载者:
相关文件:
收藏者: