cvnet 0.1.1

Fast Use Computer Vision Tools

# cvnet Build Model for Computer Vision(CV) Neural Network.

## 图像分类

## 图像分割

• 语义分割

• 实例分割

• 全景分割

### 技术演化路径

1. 2010年前，传统分割：1）边缘检测；2）遗传算法

2. 2010-2015年，机器学习：1）随机森林；2）支持向量机

3. 2015年后，深度学习：1）经典分割算法：FCN, U-Net, SegNet, DeepLab; 2)实时分割算法：ENet, LinkNet, BiSeNet, DFANet, Light-Weight RefineNet; 3)RGB-D分割算法：RedNet, RDFNet

### Networks implemented

• [PSPNet](https://arxiv.org/abs/1612.01105) - With support for loading pretrained models w/o caffe dependency

• [ICNet](https://arxiv.org/pdf/1704.08545.pdf) - With optional batchnorm and pretrained models

• [FRRN](https://arxiv.org/abs/1611.08323) - Model A and B

• [FCN](https://arxiv.org/abs/1411.4038) - All 1 (FCN32s), 2 (FCN16s) and 3 (FCN8s) stream variants

• [U-Net](https://arxiv.org/abs/1505.04597) - With optional deconvolution and batchnorm

• [Link-Net](https://codeac29.github.io/projects/linknet/) - With multiple resnet backends

• [Segnet](https://arxiv.org/abs/1511.00561) - With Unpooling using Maxpool indices

#### Upcoming

• [E-Net](https://arxiv.org/abs/1606.02147)

• [RefineNet](https://arxiv.org/abs/1611.06612)

### DataLoaders implemented

• [CamVid](http://mi.eng.cam.ac.uk/research/projects/VideoRec/CamVid/)

• [Pascal VOC](http://host.robots.ox.ac.uk/pascal/VOC/voc2012/segexamples/index.html)

• [ADE20K](http://groups.csail.mit.edu/vision/datasets/ADE20K/)

• [MIT Scene Parsing Benchmark](http://data.csail.mit.edu/places/ADEchallenge/ADEChallengeData2016.zip)

• [Cityscapes](https://www.cityscapes-dataset.com/)

### Demo

1. demo site: https://www.remove.bg/upload

2. 演示效果：

demo1:

<img src=”./docs/7.jpg” width=”600” />

remove background:

<img src=”./docs/7-removebg-preview.png” width=”600” />

demo2:

<img src=”./docs/red_car.png” width=”600” />

remove background:

<img src=”./docs/red_car-removebg-preview.png” width=”600” />

# Reference 1. [ClassyVision](https://github.com/facebookresearch/ClassyVision) 2. [Deep-Learning-Project-Template](https://github.com/L1aoXingyu/Deep-Learning-Project-Template) 3. [pytorch-semseg](https://github.com/meetshah1995/pytorch-semseg) 4. [torchcv](https://github.com/donnyyou/torchcv) 5. [pytorch-cnn-finetune](https://github.com/creafz/pytorch-cnn-finetune) 6. [PaddleOCR](https://github.com/PaddlePaddle/PaddleOCR)