Skip to content

I will upload many semantic segmentation models to this repository for you to learn

License

Notifications You must be signed in to change notification settings

geekzhu001/Semantic-Segmentation

 
 

Repository files navigation

Semantic-Segmentation语义分割模型在Keras当中的实现


目录

  1. 所需环境 Environment
  2. 使用方法与数据集下载 Download
  3. 训练效果 Performance
  4. 参考资料 Reference

2020/5/13更新

将所有的loss进行了修改,此前所用的binary_crossentropy不是特别好,换成categorical_crossentropy了!

所需环境

tensorflow-gpu==1.13.1
keras==2.1.5

使用方法与数据集下载

你可以下载后进入你所想要训练的模型的文件夹,然后运行train.py进行训练。
在训练之前,需要先下载数据集,并将其存储到dataset中。
大家关心的多分类的代码在Muiti_Class_deeplab_Mobile里。

斑马线数据集:
链接:https://pan.baidu.com/s/1uzwqLaCXcWe06xEXk1ROWw 提取码:pp6w

VOC数据集:
链接: https://pan.baidu.com/s/1Urh9W7XPNMF8yR67SDjAAQ 提取码: cvy2

训练效果

原图Before 原图Before

SegNet_Mobile

处理后After processing 处理后After processing

Unet_Mobile

处理后After processing 处理后After processing

pspnet_Mobile

处理后After processing 处理后After processing

pspnet_Multi_Mobile

处理后After processing 处理后After processing

Reference

image-segmentation-keras

About

I will upload many semantic segmentation models to this repository for you to learn

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%