Mar-16-2022: Based on the latest Class-aware Regularizations (CAR), CAA + ConvNeXt-Large achieved 64.12% mIOU on Pascal Context!
Some segmentation results on Flickr images:
- Install TensorFlow (>= 2.4, 2.3 is not recommend for GPU, but okay for TPU)
- Install iSeg (My personal segmentation codebase, update soon at https://github.com/edwardyehuang/iSeg)
- Install iDS (Dataset supports for iSeg, update soon at https://github.com/edwardyehuang/iDS)
- Clone this repo
Since some of the original experiments (especially for ResNet-101) are conducted a long time ago, the ckpts listed below may have slightly different performance with paper reported.
Backbone | ckpts | mIOU% | configs |
---|---|---|---|
ResNet-101 | weiyun | 55.0 | configs |
EfficientNet-B7 | weiyun | 60.3 | configs |
Backbone | ckpts | mIOU% | configs |
---|---|---|---|
ResNet-101 | weiyun | 41.2 | configs |
Backbone | ckpts | mIOU% | configs |
---|---|---|---|
EfficientNet-B5 | weiyun | 47.27 | configs |
@article{Huang2021CAAC,
title={Channelized Axial Attention -- Considering Channel Relation within Spatial Attention for Semantic Segmentation},
author={Y. Huang Di Kang and Wenjing Jia and X. He and Liu Liu},
journal={ArXiv},
year={2021},
volume={abs/2101.07434}
}