This is a Pytorch implementation of DeepLab-Xception, It's based on a Modified Aligned Xception backbone. Currently, we train DeepLab V3+ from scratch, using Pascal Voc 2012 dataset.
We use deeplab v3+ model trained on Pascal VOC 2012 and SBD datasets to inference these results. After 50 training epoch, our deeplab v3+ model can reach 72.7% mIoU on Pascal VOC 2012 test set.
Python3.x (Tested with 3.5)
PyTorch (Tested with 0.4.0)
tensorboardX
opencv-python
This code is tested in Ubuntu 16.04. To use this code, please do:
-
Clone the repo:
git clone https://github.com/jfzhang95/pytorch-deeplab-xception.git cd pytorch-deeplab-xception
-
To train DeepLabV3+, please do:
python train.py
- Basic deeplab v3+ model
- Training deeplab v3+ on SBD dataset
- Results evaluation on Pascal VOC 2012 test set
- Deeplab v3+ model using resnet as backbone
- Training deeplab v3+ on other datasets